A Republic of Equals: A Manifesto for a Just Society 9780691189987

Citation preview

A REPUBLIC OF EQUALS

JONATHAN ROTHWELL

A REPUBLIC OF EQUALS A MA N I F ES TO FOR A J U S T S OC I ET Y

PRINCETON UNIVERSIT Y PRESS PRINCETON AND OXFORD

Copyright © 2019 by Prince­ton University Press Published by Prince­ton University Press 41 William Street, Prince­ton, New Jersey 08540 6 Oxford Street, Woodstock, Oxfordshire OX20 1TR press​.­princeton​.­edu All Rights Reserved Library of Congress Control Number: 2019944499 ISBN 978-0-691-18376-3 ISBN (e-­book) 978-0-691-18998-7 British Library Cataloging-­in-­Publication Data is available Editorial: Sarah Caro, Charlie Allen, and Hannah Paul Production Editorial: Jenny Wolkowicki Jacket design: Emily Weigel Production: Erin Suydam Publicity: James Schneider and Kate Farquhar-­Thomson This book has been composed in Garamond Printed on acid-­free paper. ∞ Printed in the United States of Amer­i­ca 10 ​9 ​8 ​7 ​6 ​5 ​4 ​3 ​2 ​1

CONTENTS

Acknowl­ edgments vii

1 ­Behind the Discontent  1 2 The Natu­ral Foundations of a Just Society  51 3

Merit-­Based Egalitarianism 

68

4 The Importance of Equal Access to Public Goods and Markets  99 5 Unequal Access to Education  116 6 The Historical Contingencies of Group Differences in Skills  148 7 Unequal Access to Housing Markets  191 8 How Unequal Access to Housing Perpetuates Group In­equality and Injustice  219

9 Unequal Access to the Buying and Selling of Professional Ser­vices 

242

10 Creating a Just Society  273 Appendix 295 Notes 317 Index 369

ACKNOWL­E DGMENTS

As my ­father likes to say: “Every­one and anyone can escape the prison of their birth.” That’s one of his favorite syllogisms. His life was proof enough for him, and he maintains that it would have been impossible without the opportunities afforded by education. ­A fter a modest upbringing in an Irish-­Swedish f­ amily in Boston, he enlisted in the military, played professional hockey, earned a bachelor’s degree from a hockey scholarship, received benefits from the GI Bill, and went on to earn a doctorate and have a successful ­career in the administration of higher education; along the way, he met my m ­ other, who became a preschool teacher and devoted convert to Roman Catholicism. She grew up in suburban Philadelphia in an English-­A merican ­family who traces their American ancestors back to the 1600s. Despite very dif­f er­ent f­ amily backgrounds and personalities, ­they’ve had nearly 50 years of happy marriage. I dedicate this book to them. Their love and careful investment in my development and education made this book pos­si­ble. I’d also like to thank the many other teachers—at all levels of education—­ scholars, professors, colleagues, friends, and f­ amily who have invested in me throughout my life and tolerated and encouraged my devotion to justice. ­There are too many to thank, but I want to mention gratitude for the opportunity to study with a few who ­were particularly influential: Vincent Colapietro, Judith Van Herik, Ken Hirth, Leswin Laubscher, Roger Brooke, Duncan Foley, Anwar Shaikh, Lance Taylor, Carles Boix, Anne Case, Esteban Rossi-­Hansberg, Paul Krugman, Atul Kohli, Katherine Newman, and Jesse Rothstein, among other excellent and generous professors and scholars. My mentor, advisor, and co-­author Douglas S. Massey has been particularly influential.

viii • Acknowl­ edgments

I am also extremely grateful to Mark Muro as well as Alan Berube, Amy Liu, and Bruce Katz at The Brookings Institution for hiring me straight out of gradu­ate school, recognizing something in my potential and fostering it over seven years. Fellow Brookings colleagues Richard Reeves, Carol Graham, Neil Ruiz, and Isabelle Sawhill w ­ ere also encouraging and helpful. I owe a ­great debt to Andrew Reamer and ­others at the George Washington University Institute of Public Policy for their willingness to welcome me as a visiting scholar and give me access to their library resources. My Gallup colleagues, especially Patrick Bogart and Ilana Ron-­Levy, and our leader, Jim Clifton, deserve special thanks for giving me the time and resources to work on this book and enrich the analy­sis with Gallup’s wealth of data and expertise. It would have been impossible to write this book without their support. A handful of scholars and friends generously reviewed e­ arlier drafts or pre­ sen­ta­tions and helped me make this book better than it would other­wise have been, including Carol Graham, Ashley Jardina, Carles Boix, Scott Winship, Richard Florida, Andre Perry, Richard V. Reeves, Eric Turkheimer, Dalton Conley, Piotr Paradowski, Jim Harter, Julie Ray, Xavier Harris, and Naftali Bean Rutter. Fi­nally, I am grateful and fortunate to have benefited from the support and wise editorial guidance of Sarah Caro at Prince­ton University Press, the excellent editing of Erin Davis, and the anonymous referees who reviewed early drafts of the manuscript and helped me improve it.

A REPUBLIC OF EQUALS

1 ­Behind the Discontent

“Our country is in serious trou­ble,” began Donald Trump as he formally announced his candidacy for president of the United States in June 2016. He listed China, Japan, and Mexico as aggressors who are “beating us” and “killing us eco­nom­ically” via bad trade agreements. Mexico, meanwhile, is further harming the United States through immigration, he claimed: “The U.S. has become a dumping ground for every­body ­else’s prob­lems.”1 In remarks that would be much quoted and criticized, he said: When Mexico sends its ­people, ­they’re not sending their best. ­They’re not sending you. Th ­ ey’re not sending you. ­They’re sending ­people that have lots of prob­lems, and t­ hey’re bringing ­those prob­lems with us. ­They’re bringing drugs. ­They’re bringing crime. ­They’re rapists. And some, I assume, are good ­people. Trump eventually won 46 ­percent of the vote in the 2016 U.S. presidential election. His victory followed significant gains by nationalist parties in the Eu­ro­pean Union’s 2014 parliamentary elections, and a shocking vote by the United Kingdom to leave the Eu­ro­pean Union.2 In 2017, Marine Le Pen, representing France’s National Front party, won 21.4 ­percent of the vote in the first round of France’s presidential election before eventually losing in the second round.3 The nationalists seem to have two t­ hings in common: An insistence that their countries are declining, eco­nom­ically and culturally, and the identification of external forces as the reason—­with trade and immigration being primary suspects. ­These views are badly mistaken, but the nationalists have a point about the ill-­functioning of the economy, and much of the public shares

2 • Chapter 1

their sense that something impor­tant is wrong with their country’s po­liti­cal leadership. This chapter lays out what is wrong and why. Rising income in­equality and slow economic growth have been two of the most striking patterns in rich countries during the last 35 years. The explanation is not trade or technological innovation; nor is it mass migration or the rise of global superstars. Rather, countries are becoming more inefficient and unequal b­ ecause services—­ which are regulated and controlled by elite associations to the benefit of their members—­are taking over the economy, and a small group of elite ser­vice providers has managed to secure much of the gains for itself via the gradual accumulation of rights and privileges that elevate this group above markets.

VOTES OF NO CONFIDENCE

In most of the world’s richest countries, po­liti­cal discontent reigns. Confidence in government is low and has fallen steadily in recent years. In 2006, 43 ­percent of residents living in Organisation for Economic Co-­Operation and Development (OECD) member countries—­the world’s 35 richest democracies—­ expressed confidence in their national government, when asked by the Gallup World Poll.4 By 2016, that already low share had fallen to just 37 ­percent of residents. Shockingly, that’s lower than the global average of 54 ­percent. Confidence has plummeted in a number of countries that have seen a rise in support for nationalist parties or politicians, including Greece, Finland, the United States, Denmark, the United Kingdom, and Austria (figure 1.1).5 Confidence, which was already low, also fell in France ahead of a strong second place finish by the far-­right National Front in the presidential election. Just 28 ­percent of French residents expressed confidence in 2016. As of this writing (early 2019), a “yellow jacket” populist movement has upended French politics with massive street protests in response to rising taxes on diesel, and according to reports, rising housing and living expenses.6 In the United States, confidence is only slightly higher at 30 ­percent. On the other hand, low or declining confidence in national government is not inevitable in rich countries. In the Netherlands, confidence increased from 43  ­percent to 57  ­percent. In Switzerland, confidence went from 63 percent to 80 ­percent. It is also up and relatively high in Canada and Germany. Extreme po­liti­cal parties have not been as successful in t­ hese coun-

­Behind the Disconten • 3 Chile Greece Finland United States Spain Slovenia Denmark Mexico Belgium Portugal Estonia United Kingdom Australia Austria New Zealand Ireland Hungary France Norway –40% –35% –30% –25% –20% –15% –10% –5%

0

Figure 1.1. ​OECD countries with declining confidence in national government, 2006–2016 Source: Gallup World Poll, via Gallup Analytics. Chart shows change in the percentage of residents who expressed confidence in their government in 2016 compared to 2006.

tries, with the exception of Germany, where the far-­right Alternative for Germany Party (AfD) has gained traction, reportedly, in response to the governing party’s ac­cep­tance of a large number of refugees.7

SLOW ECONOMIC GROWTH, SPREAD LESS EVENLY

Lurking ­behind the rising discontent has been a major international slowdown in economic growth, or the rate at which living standards increase. The financial crisis that originated in a U.S. housing market ­bubble certainly

4 • Chapter 1 4.5% 4.0% 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

0.0%

OECD GDP per worker growth

U.S. GDP per worker growth

Figure 1.2. ​The 50-year slowdown in productivity growth in OECD countries compared to the United States, 1964–2014 Source: Penn World Tables 9.0, Real GDP divided by employment, annualized growth over ten-year periods.

had a large negative effect, but the growth slowdown preceded that, and in many ways, the housing ­bubble can be understood as a desperate attempt to find profitable investments in a low-­growth world. In 28 of the 30 OECD countries with available data, the rate of growth has been slower from 1980 to 2014 than from 1960 to 1980. New Zealand and Sweden ­were the only rich democracies to avoid this fate. For the average resident of an OECD country, the annual growth rate in gross domestic product (GDP) per capita slowed from 3.8 ­percent between 1960 and 1980 to 2.1 ­percent between 1980 and 2014. This ­can’t be attributed to the aging baby boomers dropping out of the l­abor force. Growth has also slowed on a per worker basis, from 3.5 ­percent to 1.9 ­percent (figure 1.2).8 From 2008 to 2014, annual growth has been particularly weak, just above 1 ­percent in ­either per capita or per worker terms. Productivity growth—­the fundamental source of long-­run living standards—­has slowed. This means an entire generation has now come of age in a less dynamic society than the one experienced by its parent’s generation. Partly b­ ecause of the G ­ reat Recession, growth since 2008 has been particularly weak, but the slowdown started even before then.

­Behind the Disconten • 5 United States United Kingdom Canada Poland Germany Korea Ireland Switzerland France Japan Portugal Hungary Italy Czech Republic Australia Sweden Spain New Zealand Norway Finland Denmark Netherlands 0%

20.2% 13.9% 13.6% 13.3% 13.2% 12.2% 11.5% 11.3% 10.8% 10.4% 9.7% 9.6% 9.3% 9.2% 9.1% 8.7% 8.5% 8.1% 7.8% 7.4% 6.4% 6.3% 5%

10%

2014 (or latest available)

15%

20%

25%

1980 (or closest available)

Figure 1.3. ​Share of national income ­going to richest one ­percent in 2014 and 1980 Source: World In­equality Database, accessed October 2018 via STATA program WID.

Slow growth, it must be said, is not a disaster. It is much better than no increase in living standards or, even worse, a decline. For most rich countries, living standards have continued to improve. Yet, this minimal pro­gress has been undermined by a disturbing trend, famously documented by the French economist Thomas Piketty and his collaborators. The ­earlier robust round of growth—­a fter World War II—­ occurred ­under conditions of falling income in­equality, while the latest round has coincided with rising income in­equality. From 1980 to 2014, the richest 1 ­percent of taxpayers took home a larger share of national income in ­every rich country with comparable data (figure 1.3).

6 • Chapter 1

Modest growth, combined with high in­equality, results in ­little to no gains for substantial portions of the population. This can be seen for the United States, which, by all accounts, stands out as one of the most unequal developed countries in the world and has become more unequal over the last 40 years. The economists Piketty, Saez, and Zucman have used tax rec­ords and other public sources to trace where income growth has gone in the United States. From 1980 to 2014, the share of income ­going to the top one ­percent doubled from 10 to 20 ­percent, while the bottom 50 ­percent of taxpayers (­those with taxable income at or below the median) saw their share fall from 20 ­percent to 13 ­percent. Average pretax real income for this group increased by only 1 ­percent from 1980 to 2014.9 The ­middle distribution (­those between the 50th and 90th percentiles) saw a 42 ­percent increase, but most of the gains went to the top 10 ­percent of income earners.10 Overall, the weak growth in income for t­ hose outside the very top made it difficult to accumulate wealth, according to the data from Piketty and collaborators. In 1980, the bottom half of the distribution had a meager $3,405 in average wealth (in 2014 USD).11 Wealth for this group peaked in 1992 around $7,000 and declined thereafter—­even turning negative during the ­Great Recession. As of 2014, average wealth was just $349 for this group. More recent data from the Survey of Consumer Finances—­analyzed by economist Edward Wolff—­shows the same pattern: zero increase in wealth for h ­ ouse­holds in the ­middle of the wealth distribution. For the median U.S. ­house­hold, wealth was $78,000 in 2016 compared to $80,000 in 1983, in inflation-­adjusted dollars.12 Even as overall income growth stagnated for the bottom half of earners, ­there ­were notable changes across income sources. The weak positive growth ­didn’t come from taxable ­labor income—­which is the kind of income that shows up in bank accounts and can be readily used for spending. That fell by nearly 10 ­percent from 1980 to 2014 for the bottom 50 ­percent of U.S. adults. ­These losses ­were offset by gains in tax-­exempt ­labor income—­from healthcare and retirement benefits—­with modest growth from capital gains associated with home owner­ship.13 Thomas Piketty, Emanuel Saez, and Gabriel Zucman put it directly: “The bottom half of the adult population has thus been shut off from economic growth for over 40  years, and the modest increase in their post-­tax income has been absorbed by increased health spending.”14 The falling share of income ­going to this group has impor­tant implications for living standards. If their 1980 income share and economic growth had

­Behind the Disconten • 7

both been unchanged—­a big assumption, no doubt—­then the bottom 50 ­percent of Americans would have earned $26,000 on average in 2014, instead of only $16,000.15 ­Here’s another way to think about the scale of the change in in­equality: Divide the U.S. adult population in half. Tell the poorest half that they all need to pay $10,000 ­every year to the richest 1 ­percent of Americans, ­those with annual taxable incomes of roughly $300,000 or more. For many reasons, the same ­people ­won’t be on the top or bottom e­ very year (e.g., income typically rises with age u­ ntil just before retirement). Still, regardless of who they are in any given year, the rich and poor used to be much closer together, and economic growth used to raise living standards more evenly across income groups. Indeed, Piketty, Saez, and Zucman’s data show that income growth was higher for lower income groups than it was for the rich in the de­cades preceding 1980. The basic point is ­simple: Economic growth is less effective at raising the living standards of the masses, when income in­equality is high—­and a small slice of elites reaps a huge chunk of the gains. That is likely to strike many ­people as unfair. Accordingly, levels of confidence in national government could be explained in large part by looking at both growth and income in­equality. Between 2006 and 2016, confidence did not increase in any of the 16 OECD countries that score below average on recent growth (figure 1.4). The places with low growth and high in­equality saw the largest drop in confidence (15 percentage points on average). The United Kingdom’s Brexit vote occurred ­under slow-­growth/high-in­equality conditions. However, it was also the case that poor per­for­mance on ­either in­equality or growth was generally associated with flagging confidence in government. Thus, confidence fell in France during a period of low growth and low in­equality, whereas Donald Trump was elected president of the United States with above average growth and in­equality. Meanwhile, confidence in government r­ ose in countries with high growth and low in­equality by 5 percentage points. This describes Sweden and Germany. At the individual level, ­there is also a very strong correlation between income, financial security, and feelings ­toward government within OECD countries. When it comes to predicting their confidence in government or approval of national leaders, p­ eople’s feelings about their income are far more impor­tant than their a­ ctual income or even rank within society. Gallup World Poll has several items that capture financial anxiety, including ­whether economic conditions in the respondent’s local area are getting better or worse and

8 • Chapter 1 5.0%

–2.9% –6.0%

–15.4% Low growth, high inequality

Low growth, low inequality

high growth, high inequality

high growth, low inequality

Figure 1.4. ​Change in confidence in national government, shown by groups of countries defined by levels of economic growth and inequality, 2006–2016 Note: Growth is measured using GDP per capita from 2008 to 2015; inequality uses the most recent measure of income inequality. Source: Gallup World Poll, via Gallup Analytics, World Bank, and All the Ginis (ATG).

­ hether or not the respondent is living comfortably off his or her ­house­hold w income or having difficulty making ends meet. The relative income of an individual’s ­house­hold has a small effect in the expected direction. Lower-­income individuals are a bit less satisfied with the national government and its specific leader: top-­quintile h ­ ouse­holds hold confidence or approval ratings roughly four percentage points higher than bottom-­quintile ­house­holds. ­Those with lower education also tend to be somewhat less confident in their governments. Yet, income and education are incomplete mea­sures of financial security. Subjective judgments pick up nuances that are specific to an individual’s life circumstances, including wealth, debts, the strength of supportive f­ amily ties, expectations for the near f­ uture, and po­liti­cal views that may bias their interpretation of the economy. More-­anxious individuals are much less satisfied with government. A standard deviation increase in financial anxiety predicts a 16 and 18 percentage point drop in approval and confidence, respectively.16 The upshot is that even middle-­class and affluent h ­ ouse­holds are likely to turn against a government when they feel their financial conditions are becoming difficult, particularly if ­others seem to be gaining at their expense and they are dissatisfied with their living standard. Rising in­equality and slow growth make that dissatisfaction more likely.

­Behind the Disconten • 9

I agree with a number of scholars who have argued that slow growth in an unequal society is corrosive to politics, but the details of the argument are impor­tant.17 Rising in­equality in a growing society might come from super-­ star innovators, entrepreneurs, and performers, and one might argue they deserve to be rich, b­ ecause their skills generate value for every­one. Yet, while that has happened, it does not accurately describe the last several de­cades, which have been characterized by low growth. The a­ ctual circumstances of weak growth and rising disparity indicate that something is fundamentally out of sorts. It may be that the pro­cess that leads to slow growth is making a  small elite rich and hindering opportunities for a wide swath of the population. Thus, figuring out the c­ auses of both slow growth and rising in­equality has enormous implications for what, if anything, can be done about them.

THE NATIONALIST EXPLANATIONS: BLAME THE FOREIGNERS

Nationalists on the Right, like Donald Trump, blame trade and immigration for impoverishing the m ­ iddle class. By nationalists, I mean t­ hose who view the world as a conflict between citizens of a national community—­usually thought of as ­those who have ancestral roots ­going back at least several generations in a country—­and every­one ­else. Trump and some of his allies are also populists in the sense that they view elites—­whether in business, culture, or academia—as hostile to common citizens. By contrast, left-­wing populists, like U.S. senator Bernie Sanders, evince similar skepticism ­toward trade and immigration and likewise view elites (cap­i­tal­ists) as acting against the interests of common citizens (workers), but do not suggest the conflict is rooted in ethnic or national identity, and so are not nationalists. I have some sympathy with populist complaints about elite influence on the economy, as the book ­will discuss, but I believe that both of ­these perspectives fundamentally misidentify the core prob­lems. Start with the populists and trade. Across rich countries, the share of workers employed in the manufacturing sector has declined for two reasons: automation/machines do much of the work once performed by ­human hands, and much of the work that still is more efficiently done by ­humans can be done far more cheaply in developing or middle-­income countries like China, Poland, Mexico, and Turkey.

10 • Chapter 1

Trade is certainly an impor­tant contributor to industrial decline in developed countries. By comparing employment changes across counties and networked groups of counties (commuting zones), economists have estimated that trade with China is responsible for roughly one-­quarter of the five million net job losses in the U.S. manufacturing sector from 1991 to 2011.18 ­These losses also appear to have led to a decline in the incomes of manufacturing workers, as they took pay cuts or worked fewer hours. For the average American manufacturing worker, the best available estimate of the impact of trade competition on individual wages for workers in highly exposed industries is a reduction of just over $1,000 per year over the period from 1992 to 2007.19 Yet, the news i­sn’t all bad. Chinese imports, which ramped up dramatically in the 2000s, did not coincide with a reduced risk of unemployment for manufacturing sector workers relative to workers in other industries that ­were not exposed to import competition but did face intense domestic competition. By some measures—­like risk of layoff or short job tenure—it was actually safer to be a manufacturing worker over the last 15 years than a worker in retail, construction, or even professional ser­vices.20 More importantly, only about one in 20 Americans work in manufacturing and even fewer in the industries that compete directly with China. The vast majority of working-­and middle-­class Americans have benefited from cheaper imports, often disproportionately, ­because Chinese imports tend to compete on the mass market—­think Wal-­Mart—­rather than on the market for luxury goods. On balance, it’s likely that imports from China have lowered income in­equality by disproportionately boosting the spending power of lower-income Americans, as Christian Brodis and John Romalis concluded ­after a careful study of consumption data, which proves that trade with China raised the spending power of the U.S. working class—­and, to a ­great extent, all U.S. residents.21 The same point can be made by looking at consumption data by income group from the Bureau of ­Labor Statistics. Families in the ­middle of the income distribution spent more than their after-­tax income, which implies they borrowed money to meet expenses. Meanwhile, families in the 90th percentile saved 24 ­percent of their after-­tax income and so benefited proportionately less than lower-­income families from falling prices for goods. It is easy to see how this connects to trade. Middle-­income families spend 25 ­percent of their after-­tax income on food, clothing, and automobiles.22 Families in the top 90th percentile spend just 15 ­percent of their income on t­ hese t­ hings.

­Behind the Disconten • 11

­ ese goods—­along with computers and cell phones, which have seen massive Th price declines—­are precisely the industries most exposed to import competition. In this way, mass imports have resulted in a progressive re­distribution of purchasing power. Without imports, in­equality would be worse. The economic effects of immigration are more complicated, but t­ here are no logical or empirical grounds for linking immigration to the rising share of income going to the rich. In general, immigrant l­abor has prob­ably resulted in lower incomes for ­people working in the occupations most exposed to immigrant competition—­ such as construction workers, restaurant workers, cleaners, and childcare workers—­though even this finding is disputed with strong empirical evidence from specific cases assembled by Michael Clemens and Jennifer Hunt, for example.23 In any case, economists tend to agree that most Americans—­including most of the m ­ iddle and working classes—­have seen net benefits, and that is the view of a recent comprehensive analy­sis from a committee of experts at the National Academies.24 The benefits of immigration come primarily from two channels: greater spending from immigrants, which stimulates local demand for goods and ser­vices, and lower prices for ser­vices provided by immigrant entrepreneurs and workers. Immigrants, of course, also frequently start businesses and contribute to innovation, culture, and scientific advances; t­ here the benefits are harder to quantify but are nonetheless impor­tant. In t­ hese ways, all but a small segment of Amer­i­ca’s ­middle and working classes have been enriched, at least somewhat, through mass immigration. The selective immigration of highly educated workers and entrepreneurs—­through student or skills visas, for example—­has been an even more obvious benefit to the average American. The downsides of mass migration are not trivial but are not likely to have led to the rise in top income shares. George Borjas, an economist at Harvard, has assembled evidence over de­cades of research showing that mass immigration has lowered wages of other immigrants and U.S.-­born workers with less than a high school diploma—­a small share of the population. That finding has been disputed, but it, at least, has clear logic b­ ehind it in that employers can reduce wages when many new workers enter the l­ abor market.25 Even so, the estimated effects are small.26 The large wage gap between elite professional workers or business ­owners and ­those with less than a high school diploma cannot be explained by immigration, and most estimates suggest that U.S. workers with moderate levels of education saw long-­run wage benefits from mass immigration.

12 • Chapter 1

Aside from creating wage pressure on other less educated workers, many immigrant workers with low levels of education consume more in government benefits than they pay in taxes over their lifetimes, creating fiscal bud­getary challenges.27 Yet, one could say the same about U.S.-­born Americans with low levels of education, as the National Academies report shows.28 The fact is that undereducated immigrants are a lower fiscal burden than undereducated U.S.-­born residents. For both populations, the net effect reflects, in part, the fact that the U.S. government typically runs a large bud­get deficit for po­liti­cal reasons entirely unrelated to immigration. If one makes the reasonable assumption that an additional immigrant does not increase the costs of pure public goods like national defense, then the average immigrant—­ especially ­those with higher educational levels—­has had a net positive effect on government finances. The favorable long-­term effects of immigration are consistent with the most comprehensive analy­sis ever of intergenerational mobility. Using data from millions of tax rec­ords and the Census, economists from Harvard and the Census Bureau found that Hispanics living in the United States have high rates of relative intergenerational income mobility, such that incomes are converging with non-­Hispanic white p­ eople.29 The c­ hildren of Asian immigrants, meanwhile, are earning more as adults than non-­Hispanic white ­children raised with similar ­family incomes. Neither an increase in immigrants from Latin Amer­i­ca nor an increase in imports from China can explain the rise of incomes at the top. As we w ­ ill see, ­people in the manufacturing sector—­including the executives who would have benefited the most from offshoring production—­make up a tiny fraction of the one p­ ercent. Likewise, immigrants lacking college degrees rarely comprise a substantial share of workers at the international corporations that pay top executives, nor at hedge funds, law firms, and physicians’ offices. Data across rich democracies tell the same story: OECD countries that are more exposed to immigration or trade are no more unequal and no more likely to have become more unequal since 1980. W ­ hether one mea­sures exposure to immigration as the foreign-­born share of population in 2015 or the change in the foreign-­born share from 1990 to 2015, neither is significantly correlated with the level and change in the top one ­percent’s income shares. Trade exposure can be mea­sured in a variety of ways (e.g., average tariff rates, the trade balance as a share of GDP, changes in the trade balance, or a “trade freedom index”), but none of them is even moderately correlated with the level

­Behind the Disconten • 13

or change in the top one p­ ercent’s income shares, except that countries with higher tariff rates—­and hence less exposure to globalization—­have significantly higher in­equality.30

THE CONVENTIONAL LEFT-­L EANING VIEW: IT’S CORPORATE GREED

U.S. senator Bernie Sanders ran for president in 2016 and nearly won the Demo­cratic Party primary. As a self-­described demo­cratic socialist, Sanders is concerned about income in­equality and documents the rise of in­equality in the United States on his campaign website.31 It’s the top issue on his list, and on his list of policy solutions, the first starts as follows: “Demanding that the wealthy and large corporations pay their fair share in taxes.” In fact, other than changing trade deals, his primary solutions for reducing in­equality seem to focus on three ideas: higher taxes on corporate profits, an increased minimum wage, and encouraging workers to join ­unions. At bottom, this view has one explanation in mind for the rise of the top one p­ ercent in the United States and other countries: increasing corporate power. Sanders and his aides are far from alone in embracing this view and ­these solutions to combating in­equality. The argument goes that working people lost considerable bargaining power as u­ nions declined in influence over the last several decades—­a decline that coincided with large losses in manufacturing jobs, a fall in income tax rates for the rich and for corporations, and an increasing shift in U.S. income to corporate profits and capital. It follows that ­unions and the federal government are needed to forcefully combat rising corporate power through taxation, minimum wage requirements, and perhaps other regulations. In academic circles, the ­lawyer and former U.S. ­labor secretary Robert Reich has been influential in embracing this view, and he and others have emphasized what is known as the ­labor share of income. All rich countries keep national accounts. Using a combination of mandatory surveys and tax rec­ords, statistical offices keep track of where the nation’s money comes from. GDP can be mea­sured as the sum of the nation’s income. Take ­every resident’s salary, wage, business, and investment income and you essentially have GDP.

14 • Chapter 1

Since the beginning of the field, economists have been almost obsessed with what share of income is g­ oing to l­ abor—by which they mean wage and salary income—­versus capital—­the income earned from asset owner­ship. Across most countries, the share of national income coming from wages and salaries has declined, and this has led many economists—­such as ­those mentioned above—to conclude that workers are being exploited by corporations.32 A search of Google Scholar for “­labor share of income” finds that ­there have been 1,480 academic articles discussing the topic since 2000. Writing in the New York Times, Jared Bern­stein, the left-­leaning economist and former aide to Vice President Joe Biden, argues that the under­lying cause is a fall in the bargaining power of workers.33 The International L ­ abor Organ­ization lists technology, globalization, and u­ nion power, and, more plausibly, increasing reliance on finance.34 ­There is a logical and factual basis to this perspective, but ­there are crucial flaws in the corporate power analyses. First, the capital share of income is nothing like an index of corporate power and worker exploitation. It can change for many reasons. Older ­people, for example, rely much more heavi­ly on capital income, b­ ecause many are no longer working, and most rich countries are seeing an aging population. An increase in the capital share of income follows necessarily from an increase in the share of the population that has retired and now lives, at least partly, off savings accrued during their c­ areer. The share of U.S. residents aged 65 and older increased from 9 ­percent to 11 ­percent from 1960 to 1980 and reached 13 ­percent in 2010.35 Many of ­these retired workers, who often spent their ­careers in blue-­collar jobs, would be baffled to learn that their u­ nion pension income qualifies them as “cap­i­tal­ists” by left-­wing writers or politicians. The OECD compiled data on the world’s largest pension funds. Many of them w ­ ere U.S.-­based u­ nions. CalPERs manages the retirement accounts of the California state and local government workers. As of 2014, it had $296 billion in assets u­ nder management, ­because California is a very large state with many retired public employees. The New York State Combined Retirement System had another $177 billion ­under management.36 Private sector pension funds can be almost as large. The General Motors pension fund has roughly $100 billion in assets.37 It would be ironic to conflate the value of ­union retirement funds with “corporate power.” Housing is another big source of capital income. Economist Matthew Rognlie has done painstaking work reassessing the claims that the capital

­Behind the Disconten • 15

share of income has increased. He concludes: “The long-­term increase in capital’s net share of income in large developed countries has consisted entirely of housing. Outside of housing, capital’s rise in recent de­cades has merely reversed a substantial ­earlier fall.”38 In ­every OECD country except Switzerland and Germany, t­ here are more homeowners than renters.39 The rise of housing income is not equally distributed, and rising housing prices are a major prob­lem for the ­middle class and poor, but this is very dif­fer­ent from a rise in capital income resulting from business owner­ship. Thoughtful analy­sis by economists Josh Bivens and Larry Mishel at the left-­ leaning Economic Policy Institute makes a similar point: “An analy­sis of income shares tells us l­ ittle about the bargaining position of l­ abor vis-­à-­vis ­owners of capital.” 40 A second flaw in the corporate power analyses has to do with the fact that the overall contribution of corporations to the U.S. economy has unambiguously declined during the era of rising in­equality. Income from corporations was over 60 ­percent of national income in the 1980s and 1990s. In the twenty-­ first ­century, corporate income has accounted for just 56 ­percent of national income. At the same time, corporate profits went from 9.2 ­percent of national income in 1980 to 12.5 ­percent in 2017 (figure 1.5). Roughly one-­third of this increase was driven by profits earned by foreign establishments of U.S. corporations.41 The domestic operations of U.S. corporations increased their share of national income by only two percentage points from 1980 to 2017, and that share is well below the contribution of corporate profits in the 1960s and 1970s, when ­unionization rates ­were much higher and the income distribution more equal. Considering that the top 1 ­percent of income earners increased their share of income by 10 percentage points since 1980, corporate profits—­whether foreign or domestic in origin—­could not have accounted for more than a third of the increase, even if all the profits went to the top one ­percent. In fact, 64 ­percent of income from stocks goes to the top one ­percent.42 Third, as we ­will see, the sectors that generate the vast majority of top earners—­including finance, healthcare, and professional services—­were never meaningfully u­ nionized, so the decline in unionization—­largely a manufacturing phenomenon—­could not have any direct effect on the incomes of most top earners. Fourth, while capital income disproportionately goes to the rich and very rich, during the most impor­tant period of rising income in­equality in the United States—1980 to 2000, when 80 ­percent of the top one p­ ercent’s gains

16 • Chapter 1 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 1948 1951 1954 1957 1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 2017

0

Domestic corporate profit share of income Domestic plus foreign profits share

Foreign corporate profits share

Figure 1.5. ​Corporate profits by domestic and foreign status as a share of national income, 1948–2017 Source: Author analysis of U.S. Bureau of Economic Analysis, “Table 1.13. National Income by Sector, Legal Form of Organization, and Type of Income,” https://www.bea.gov/ (accessed June 20, 2019).

occurred—­most of the gains to the top one p­ ercent (70 ­percent) came from ­labor compensation, not capital income, according to evidence compiled by Piketty, Saez, and Zucman.43 While capital income has contributed to rising income shares of the top one ­percent during the early twenty-­first ­century, it increasingly goes to homeowners and noncorporate businesses, which include entities such as hedge funds, law firms, physician’s offices, and hospitals. While capital income accounts for 56 ­percent of the income ­going to the top one p­ ercent, just 25  ­percent of total top one-­percent income comes from stocks as of 2014, which is down slightly from 1980.44 Fifth, tax policies ­won’t solve the in­equality prob­lem. I agree with Senator Sanders that the rich have seen a decrease in tax rates in recent de­cades, and I agree with him that increasing the progressivity of tax rates would benefit lower-­income groups and be a reasonable way to rebalance federal bud­get obligations, but it is difficult to find empirical evidence that pretax income in­equality r­ ose principally as a result of declining tax rates. Top marginal tax rates on personal income w ­ ere very high a­ fter World War II, but started declining in the 1960s, long before the rise in in­equality. It’s true that tax rates fell still further in the 1980s, but they went up again in the 1990s, even as

­Behind the Disconten • 17

l­ abor income became more unequally distributed. A progressive income tax is, I believe, a fair way to distribute the burden of paying for government ser­ vices, but it w ­ on’t solve the more fundamental issues of differences in economic opportunity and access to markets.45 Fi­nally, the corporate power story also risks exaggerating the importance of the mega-­rich. To be at the 90th  percentile of U.S. income earners in 2014, you needed $124,000 in total pretax income (taxable plus tax exempt) or $88,600 to be in the top 10 ­percent in ­labor income. To be in the top one ­percent, you needed $477,500 in total income or $268,937 in l­ abor income. To be in the top 0.1 ­percent, you needed $2 million in total income or $0.95 million in ­labor income. The top ten p­ ercent gained 12.8 percentage points of total income from 1980 to 2014. Of all that money 25 ­percent went to ­those in the top 90 to 99 ­percent (so not even the top one p­ ercent), 30 ­percent went to ­those in the top 99 to 99.9 ­percent, and 45 ­percent went to the top 99.9 ­percent. A large number of elite professionals have salaries that fall in between $88,600 and $950,000, and many of them own publicly traded stocks and noncorporate businesses. Most of U.S. income growth since 1980 went to this group.46 ­Don’t get me wrong—­the mega-­rich have contributed greatly to rising income in­equality, but so did the upper m ­ iddle class, as my friend Richard Reeves of The Brookings Institution has argued, and at least some of the mega-­rich, like the superstar entrepreneurs, have generated value for large numbers of ­people.47

THE CONVENTIONAL RIGHT-­L EANING VIEW: TO EVERY­O NE WHO HAS, MORE ­W ILL BE GIVEN

A widely accepted view among academic economists is that the top one ­percent’s growing share of income is largely a result of the way skills interact with technology. In one version of the theory, p­ eople with high skills have always been paid more ­because they are more productive, but information technology has magnified the value of complex cognitive skills (meaning job tasks that involve specialized knowledge, creativity, and abstract thinking) and lowered the value of performing routine physical or clerical work. The reason is that workers who perform complex tasks use technology in ways that increase their productivity and make them more attractive to businesses, whereas some subset of workers who perform routine tasks have had to

18 • Chapter 1

compete against automated software programs, industrial machines, and user-­friendly internet platforms. David Autor has contributed many influential papers to this theory, and in a recent summary of lit­er­a­ture, he suggests that the theory is particularly relevant for explaining income differences among the bottom 99 ­percent of income earners. The evidence for this view includes the well-­established fact that the earnings advantage of holding a tertiary degree (at the bachelor’s level in the United States) relative to having only a high school diploma grew sharply from 1980 to the pre­sent, reaching around 80 ­percent in 2010 from just over 40 ­percent in 1980.48 A related line of scholarship argues that superstar companies and individuals benefit disproportionately from expanded access to global markets. Information technology has lowered the costs of transmitting information—­ including ideas and intellectual property—­which, in turn, has expanded the market for ideas and enlarged the rewards. A blockbuster book, movie, or architectural design is worth more now than it used to be ­because content producers can sell around the world to a much greater extent than in ­earlier de­cades. This helps explain why certain businesses and individuals can become enormously rich by gaining just a fraction of global market share.49 At the heart of this research is the notion that in­equality is largely about who does and does not possess valuable skills—or “­human capital.” As to where h ­ uman capital comes from, a large lit­er­a­ture in economics holds that ­human capital comes about through investments in education and is not preordained by one’s DNA.50 Yet, Harvard economist Greg Mankiw argues that the earnings ­going to the top one p­ ercent are largely a result of their ge­ ne­tic endowment, and thus, he suggests that the large earnings of the rich may be justified by their contribution to society.51 Yet, what this theory misses is that p­ eople with not only identical credentials but also identical traits associated with earning power—­such as cognitive and noncognitive skills—­often earn wildly dif­fer­ent salaries, and not as a result of luck, but as a result of working in a specific industry or occupation that over-­rewards or under-­rewards pay for what are ultimately po­liti­cal reasons related to how institutions affect competition and rewards in specific markets. For example, the average earnings advantage ­people get from working in the financial or ­legal industry compared to working in the restaurant industry is almost as large as the college premium, even ­after accounting for individual talent.52 If pay differences resulting from skill was the only source of earnings differences, then the United States would be far more

­Behind the Disconten • 19

egalitarian, as I argue in chapter 3 in detail. For now, I simply want to lay out what evidence we would expect to see if talent alone explained the rise of the one ­percent and the cross-­country and U.S. patterns. If the theory linking talent to in­equality is right, the richest one p­ ercent should be largely and increasingly comprised of ­people who would most benefit from technological change and globalization. Some of the obvious examples include the found­ers of massively successful tech start-­ups. Certainly, Bill Gates (Microsoft), Steve Jobs (Apple), Marc Andreessen (founder of Netscape and the venture capital firm Andreessen-­ Horowitz), Mark Zuckerberg (Facebook), Jeff Bezos (Amazon), and Sergey Brin and Larry Page (both Google) are among the richest p­ eople in the world, and their companies would not have been pos­si­ble but for recent scientific and business innovations related to communications technologies. I have no doubt that t­ hese found­ers possess a g­ reat deal of skill and talent, and yet t­ here are roughly 1.7 million individuals in the top one ­percent of the U.S. income distribution. Th ­ ere are less than 4,000 domestic publicly traded companies, a number that has been declining in recent de­cades, along with the number of start-­ups.53 Likewise, globalization has vastly expanded markets for professional entertainers such as superstar actors, actresses, movie directors, models, musicians, writers, and athletes. Beyoncé and Bono sell ­music around the world and are paid a lot of money to perform internationally. No doubt, they are super-­talented. The same could be said about global tennis and soccer stars such as Serena Williams, Roger Federer, and Cristiano Ronaldo. The implication is that if you add global individual superstars to executives, high-­level man­ag­ers, and employees of superstar multinational companies, you get the one ­percent. The theory seems to suggest a merit-­based view of income in­equality—­one that may lead some ­people to feel better about the situation, or if not, to strongly support policies that create a more egalitarian distribution of skills, which I discuss in ­later chapters.54 The merit-­based interpretation that Greg Mankiw drifts ­toward reminds me of Voltaire’s character Dr. Pangloss, who was a spoof of leading intellectuals of his day: “All is for the best in the best of all pos­si­ble worlds.” ­Those outside the top one ­percent ­don’t have to accept this story. The superstar theory is relevant for explaining a small number of p­ eople in the one ­percent, but it fails even to come close to explaining the level of in­ equality in the United States or the trend t­ oward greater in­equality. Trade,

20 • Chapter 1

technology, and talent play relatively minor roles in accounting for the extraordinary income in­equality observed in the United States. If the rich consisted only of successful entrepreneurs and entertainers, ­there would be far less in­equality.

THE WORK OF THE ONE P ­ ERCENT

One ­simple way to understand why the one ­percent earn so much is to observe how they earn their money. Work can be classified into ­either an occupation—­what you do—or an industry—­what the business (or entity) you work for produces. Both are impor­tant for testing the global superstar theory of the one ­percent. Let’s start with the a­ ctual superstars in the United States. In 2015, just 0.2 ­percent of U.S. adults (1.9 ­percent of top earners) worked as athletes, artists, writers, actors, directors, coaches of professional sports teams, news anchors, fashion designers, or any other occupation broadly classified ­under arts, design, entertainment, sports, and media (­table 1.1). Since 1980, ­there has been very ­little increase in the share of ­people in the top one ­percent who come from ­these occupations. This analy­sis is based on the American Community Survey, which surveys 1 ­percent of the U.S. population, but the figures match administrative tax rec­ords analyzed by economists, one of whom works at the Internal Revenue Ser­vice.55 Accordingly, in 1979 and 2005, p­ eople in this broad occupational category comprised 1.4  ­percent and 1.7  ­percent, respectively, a trivial increase. This group of ­people may be super-­famous and super-­rich, but that is in part b­ ecause t­ here are very few of them. A key prediction of superstar theory thus fails right out of the gate. ­These results are for the United States, but international data suggest that other countries have even fewer entertainment stars among their top earners. Sufficiently detailed data w ­ ere not available for most OECD countries, but using the Luxembourg Income Study, I found comparable figures for Denmark, Estonia, Germany, France, and Israel.56 Of ­these, Denmark has the highest share of top one-­percent earners in arts, entertainment, journalism, and sports-­related occupations, and that share is just 0.8 ­percent. Thus, none of the top earners in ­these super­star entertainment occupations is in the top one ­percent in any of ­these countries.

­Behind the Disconten • 21

Table 1.1.  Share of employed U.S. residents in the top one p ­ ercent and probability of being in the top one ­percent by occupation or occupation and sector, 1980 and 2015 Share of all workers in 1%

Probability of being in 1%

1980

2015

1980

0.3% 0.1% 0.2%

0.4% 0.2% 0.3% 12%

2015

Most likely to be superstars Actors, directors, and producers Athletes, sports instructors and officials Writers and authors Chief executives and legislators

5.1% 2.7% 4.1%

2.4% 1.2% 1.5% 14%

Most likely to benefit from globalization (trade agreements   and falling transportation costs of goods) Executives and man­ag­ers working in ­manufacturing, energy, and agriculture Executives and man­ag­ers working in ­communications, transportation, and utilities Medical scientists

12%

9%

6%

4%

2%

3%

3%

3%

0.3%

0.3%

12.4%

2.1%

0.1% 0.1%

1.5% 1.1%

0.2% 0.4%

1.1% 0.6%

0.2% 0.1% 0.1% 2.8%

0.2% 0.2% 0.1% 2.8%

0.7% 1.2% 0.3% 3.4%

0.8% 3.0% 0.4% 3.0%

Most likely to benefit from information technology Computer software developers Computer systems analysts and computer scientists Electrical engineers Chemical engineers Industrial engineers Man­ag­ers and specialists in marketing, advertising, and public relations

Source: IPUMS-­USA, using data from the U.S. 1980 Decennial Census and 2015 American Community Survey.

THE UNIMPORTANCE OF TRADABLE SECTORS

The other core prediction of the superstar theory of income in­equality is that most is generated by technology-­enabled globalization.57 For superstar firms in manufacturing and technology—­including Apple, Volks­wagen, Royal Dutch Shell, British Petroleum, ­Toyota, and Microsoft—­the international market just keeps getting bigger, and global integration has resulted in lower tariffs and greater ease of ­doing business across borders. ­There have been

22 • Chapter 1

substantial reductions in tariffs over the last 20  years, according to the World Trade Organ­ization, all of which should have created more opportunities for global manufacturing and technology firms.58 And yet, if you add up all the executives, man­ag­ers, scientists, software developers, and other employees in export-­oriented sectors such as communications and manufacturing, they amount to a surprisingly small share of the one ­percent in OECD countries, just 21.6 ­percent.59 All of mining, and manufacturing, yields just 16.8 ­percent, whereas the sector that includes transportation, warehousing, and communications comprises 4.8 ­percent of OECD countries’ top income earners. Agriculture, which is often export-­oriented, employs just 2 ­percent of top earners. Stated other­wise, nearly 80 ­percent of top income earners in rich countries are in sectors that do almost no trading across international borders. Nor, as we saw, are they entertainers, media stars, writers, or athletes. For t­ hose thinking that the rise of the one p­ ercent is linked to the small but rich portion of top earners tied to global export markets, the country-­specific results put that idea to rest. Th ­ ere is a negative correlation between the share of a country’s top o­ ne p­ ercent in t­ hese export and technology-­oriented sectors and income in­equality. In fact, some of the countries with the largest increases in in­equality have a very low percentage of top earners in t­ hese sectors. ­These include the United States and Canada, where just 16 ­percent and 14 ­percent of top earners, respectively, are in mining, manufacturing, and communications (plus transportation and storage) (figure 1.6). In the Scandinavian countries, the share is above 25 ­percent, with a high of 31 ­percent in egalitarian Finland. Denmark is one of the few countries to experience almost no increase in the share of income g­ oing to the top one p­ ercent, but a relatively high share of its top earners work in export and tech-­oriented sectors.

BIG FISH IN DOMESTIC PONDS

The fact is that the majority of the one p­ ercent in rich countries work primarily in domestic industries. Th ­ ese individuals are rarely famous and globalization and information technology have had relatively l­ ittle effect on their business model or daily activities. Across OECD countries, just three broad domestic sectors account for the majority—52 ­percent—of the one p­ ercent (­table 1.2). Th ­ ese are (1) public administration, education, health, and social work; (2) real estate, renting, and

­Behind the Disconten • 23 Luxembourg Italy Switzerland Netherlands Canada United States Ireland Israel Greece France Mexico Australia Sweden Denmark Japan Germany Iceland United Kingdom Finland Hungary Poland Spain Estonia Austria

7% 8% 11% 14% 14% 16% 17% 18% 19% 20% 21% 23% 25% 26% 27% 27% 29% 30% 30% 30% 32% 38% 39% 41%

Figure 1.6. ​Share of top earners in trade-­oriented sectors, 2013 or latest available year Source: Analy­sis of Luxembourg Income Study. The sectors included are mining, quarrying, manufacturing, transportation, storage, and communications.

business activities—­including l­ egal ser­vices, and financial intermediation; and (3) finance. Some may argue that financial ser­vices are global—­and big investment banks certainly have international clients—­but exports account for just 8 ­percent of the U.S. financial sector’s value added—­compared with 51 ­percent for manufacturing.60 At the national level, Luxembourg, Ireland, the United States, and Switzerland have the highest percentages of one percenters from ­these sectors, with over two-­thirds of the one p­ ercent coming from ­these industries. Finance is particularly impor­tant in Switzerland, Luxembourg, and Ireland, but not as much in the United States and the United Kingdom, despite the internationally famous banking centers in New York and London.

24 • Chapter 1

­ able 1.2.  Share of national top one p T ­ ercent in OECD by sector Domestic sectors Public administration; education; health and social work Real estate, renting, and business activities ­W holesale and retail trade, repair; h ­ otels and restaurants Financial intermediation Construction Other ser ­vices

23% 18% 12% 11% 6% 5%

Tradable sectors Mining and quarrying; manufacturing; utilities Transport, storage, and communications Agriculture, forestry, and fishing

17% 5% 2%

Source: Luxembourg Income Study; 24 countries with data for each sector, covering 85% of the OECD’s population. Population data are from Penn World ­Tables 9.0.

Across OECD countries, but especially in ­those that are English-­speaking, healthcare, education, and public administration are the major occupations for top earners (figure 1.7). In the United States and the United Kingdom, 26 ­percent of the top one ­percent come from healthcare, education, and public administration. In Ireland, it’s 23 ­percent, in Canada, 21 ­percent, and in Australia, 19 ­percent. So ­there is a very high correlation between income in­ equality and the share of the top one p­ ercent from ­these sectors.61 Denmark, Finland, Iceland, Austria, and Japan are among the countries with relatively low shares of top earners working in ­these sectors.62 ­These professional and ­human ser­vices can also account for the extraordinary rise of the one ­percent in the United States. In 1970, just 47 ­percent of the nation’s top one p­ ercent came from t­ hese domestic ser­vice industries. By 2015, the share had risen to 66 ­percent. Most of the increase came from finance, real estate, and healthcare, but ­legal and other professional ser­vices also contributed substantially to the rise.63

WHAT TALENT, TECHNOLOGY, AND TRADE GET YOU

Having established that the one ­percent work primarily in domestic sectors that are relatively sheltered from global competition, I turn now to what they do.

­Behind the Disconten • 25 Japan Austria Finland Iceland Greece Estonia Denmark Poland Mexico Spain France Germany Italy United Kingdom Netherlands Israel Switzerland United States Ireland Luxembourg

21% 28% 33% 38% 38% 40% 46% 46% 47% 47% 48% 53% 53% 53% 63% 64% 66% 67% 71% 78%

Figure 1.7. ​Share of country’s top 1 ­percent income earners working in healthcare, education, public administration, finance, real estate, or business ser­vices Source: Analy­sis of Luxembourg Income Study.

Analysts on the po­liti­cal Left assume that the top one ­percent are largely executives from publicly traded corporations.64 In fact, executives are but a small share of the one p­ ercent. In the United States, executives and legislators comprise just 12 ­percent of the top 1 ­percent of income earners, according to data from 2015, the latest available at time of writing.65 In other OECD countries, the figures are similar, even higher. In Denmark, 25 ­percent of the one ­percent are classified as managing directors and chief executives. In Austria, 17 ­percent are classified as CEOs, se­nior officials, and legislators, and 13 ­percent are in Switzerland and Israel. In Ireland and Finland, 16 ­percent qualify; in the Netherlands, 11 ­percent; in France, 6 ­percent; in Germany, 4 ­percent; and in Luxembourg, just 3 ­percent.66 The point is that executives are a small minority of top earners in e­ very rich country. The emphasis on executives of publicly traded companies is understandable in that they are highly vis­i­ble representatives of the elite, but this focus has perpetuated the theory that capitalism itself naturally leads to

26 • Chapter 1

massive income in­equality, whereas we w ­ ill see ­there is nothing natu­ral about it. To be sure, the share is considerably larger if middle-­and lower-­level man­ag­ers—­not just executives—­are included, as they are very large in number, but this total still amounts to just over one-­third of top earners (37 ­percent) across the OECD countries.67 The United States is roughly at the OECD country average on this mea­sure in that 36 ­percent of its top one ­percent are legislators, man­ag­ers, or executives.68 In Canada, the comparable figure is on the high end at 48 ­percent.69 Overall, the share of top earners who are man­ag­ers is negatively correlated with income in­equality. As stated before, most economists and analysts on the po­liti­cal Right assume that talent, technology, and globalization explain the one ­percent, but that too is contradicted by the occupational data. In the United States, detailed data on earnings by occupation show that the largest occupational category for top earners is physicians and surgeons, who comprise 14.5 ­percent of the top one p­ ercent. The probability of being in the one p­ ercent is also highest for this group. Slightly over one out of e­ very five physicians or surgeons is in the top one p­ ercent (21.6 ­percent) compared with one out of seven for CEOs (13.9 ­percent). ­Under the techno-­talent theory of the top one p­ ercent, this s­ houldn’t be pos­si­ble. It would be difficult to think of a professional occupational group less affected by technological change and globalization than physicians. Physicians do not export their ser­vices and are so b­ ehind in their use of information technology that the U.S. government passed legislation in 2009 to fund and encourage more widespread IT adoption among doctors.70 Altogether, health professionals account for 17.7 ­percent of top earners. ­Lawyers—­who also rarely export their ser­vices and make less use of technology than many professionals and blue-­collar workers—­represent another 8 ­percent. Another 10 ­percent work in one of several financial occupations. Combined, one out of e­ very three top earners is e­ ither a health, finance, or ­legal professional. Workers in STEM occupations—­those in science, engineering, and computer programming professions—­exhibit higher cognitive scores and are more likely to use sophisticated software technologies compared with health, finance, and ­legal professionals.71 Yet, contrary to the predictions of conventional economic theory, only 3 ­percent of top earners work in computer or mathematical occupations, 2 ­percent in engineering or architecture, and

­Behind the Disconten • 27

­ able 1.3.  Occupations that represent at least 1 ­percent of the top one T ­percent, United States, 2015

Physicians and surgeons Chief executives and legislators Miscellaneous man­ag­ers ­Lawyers, judges, magistrates, and other judicial workers Financial man­ag­ers First-­line supervisors of nonretail sales workers Marketing and sales man­ag­ers Personal financial advisors Accountants and auditors Securities, commodities, and financial ser­vices sales agents Sales representatives, ­wholesale and manufacturing Management analysts General and operations man­ag­ers Miscellaneous sales representatives First-­line supervisors of retail sales workers Software developers, applications and systems software Dentists Real estate brokers and sales agents Medical and health ser­vices man­ag­ers Financial analysts

Share of total one ­percent

Probability of being in top one ­percent

14.5% 12.0% 9.0% 7.7%

21.6% 13.9% 2.7% 9.5%

4.2% 2.7% 2.5% 2.4% 2.2% 2.1%

4.5% 2.9% 3.3% 8.5% 1.5% 11.1%

1.8% 1.8% 1.5% 1.5% 1.4% 1.3%

1.7% 2.9% 2.1% 2.9% 0.6% 1.3%

1.2% 1.1% 1.1% 1.0%

10.7% 2.0% 2.1% 5.2%

Source: Analy­sis of IPUMS-­USA, 2015 American Community Survey.

1 ­percent in science and social science occupations. So the probability of being in the top one ­percent is also much lower for STEM professionals than for health, ­legal, and finance professionals. Take software development. It is arguably the most in-­demand ­career in the United States, as mea­sured by job vacancy rates, hiring difficulty, and the market value of its skills.72 Yet, just 1.3 ­percent of software developers are in the one p­ ercent (­table 1.3). A number of unglamorous professions have a higher percentage of workers in the top one p­ ercent than software developers: financial clerks (2 ­percent), financial specialists (4.1 ­percent), veterinarians (3.2 ­percent), and insurance agents (1.8 ­percent).

28 • Chapter 1

In fact, ­there ­were 1.1 million software developers working in the United States in 2015 and 140,000 dentists, but ­there ­were approximately the same number of dentists in the one p­ ercent as t­ here ­were software developers (15,000 dentists versus 15,750 software developers). Tech may be glamorous, but you are 10 times more likely to get rich g­ oing into dentistry than software development, and over 20 times more likely to be in the top one p­ ercent if you become a physician. This level of detail is difficult to get for countries outside the United States, but available data from the Luxembourg Income Study suggest that medical and l­ egal professionals comprise an impor­tant share of top earners around the world, though usually not to the same extent as in the United States. In Israel, medical doctors comprise 17 ­percent of top earners, which is close to the 15 p­ ercent figure in the United States, but in more egalitarian countries the total is usually less. In Denmark, specialized and general medical prac­ti­tion­ers represent 9 ­percent of top earners. In Germany, medical doctors comprise 6 ­percent of top earners. In Luxembourg, it’s 6 ­percent. ­Legal, social, and cultural professions—­which include l­ awyers, librarians, and entertainers—­comprise 18  ­percent of top earners in Luxembourg, 14 ­percent in Spain, roughly 10 ­percent in the United States, 9 ­percent in Ireland, 7 ­percent in Germany, and 5 ­percent in Finland, the Netherlands, and Switzerland.73 In Denmark and Estonia, l­ awyers comprise just 2 ­percent of top earners, versus 5 ­percent in Rus­sia and 8 ­percent in the United States and Israel. Meanwhile, STEM workers represent a somewhat larger share of top earners outside the United States but are still just a small fraction of top earners. Science and engineering professionals comprise 10 ­percent of top earners in Ireland, 8 ­percent in Finland and Austria, 5 ­percent in Switzerland, and 4 ­percent in the Netherlands, but just 3 ­percent in Spain and the United States. Engineers comprise 9 ­percent of top earners in France, and architects and engineers represent 9 ­percent in Israel, but just 2 ­percent in the United States.74 For ­those living in rich countries, talent, technology, and trade, as mea­ sured by cognitive ability, sophisticated computer use, and work for an export-­ oriented com­pany, usually lead to well-­paying jobs, but rarely get you to the top of the income distribution. For that you need po­liti­cal power, as we ­shall see.

­Behind the Disconten • 29 WHY PRODUCTIVITY GROWTH SLOWED

A conventional explanation for income in­equality is that the world’s most productive companies in highly competitive industries are generating enormous profits for the “winners” of globalization, while every­one ­else loses out.75 The opposite is closer to the truth. The sectors that contribute the most to the one p­ ercent contribute the least to innovation and global productivity growth. G ­ reat inventors, innovators, and cultural stars who often become rich while contributing to ­human prosperity and happiness are rare. Most top earners work in industries that are not objectively innovative. Over the last 35 years, the least productive sectors of the economy have absorbed an increasing share of productive resources while delivering fewer benefits. That explains the low productivity growth. The most impor­tant example is healthcare. In the OECD countries, healthcare spending accounted for 6.1 ­percent of total national spending in 1980, but 35 years ­later, it had grown to 10 ­percent (figure 1.8). In the United States, the healthcare share of national spending doubled from 9 ­percent to 18 ­percent over the same period. In other words, residents of the United States, through a combination of tax dollars, business spending on insurance, and personal spending, now allocate 18 out e­ very 100 dollars to healthcare. This pattern also explains the rise in income in­equality. The majority of top income earners work in financial ser­vices, real estate, l­egal ser­vices, or healthcare, and ­these sectors display slow productivity growth. Meanwhile, the highly productive sectors such as information and manufacturing have relatively few top income earners. Even advanced services—­such as computer ser­vices, software, and engineering—­constitute only a small share of top earners. What Is Productivity Growth?

Conceptually, ­there are two ways to mea­sure productivity growth. One, which is how most economists think about it, is to compare output to inputs. Output is usually mea­sured as gross value added, which can be thought of as revenue from sales once expenses paid to outside vendors are subtracted. Take a car manufacturer, for example. Its gross value added could be calculated as total vehicle sales less the amount it paid suppliers for metal, glass, tires, and

30 • Chapter 1 12 9.8%

10 7.6%

8 6

6.1%

6.7%

4 2 0

1980

1990

2000

2015

Figure 1.8. ​Share of national income spent on healthcare in OECD countries, 1980–2015 Source: OECD.STATS. “Health Care Resources: Remuneration of Health Professionals.” https://stats.oecd.org (accessed April 3, 2018).

other inputs. The easiest way to calculate productivity is to divide gross value added by the number of workers.76 While conceptually ­simple, this approach depends upon accurate price data. For example, consider a hy­po­thet­i­cal scenario in which a new law forbids the construction of new homes in a country with growing demand for housing b­ ecause of immigration and demographics. House prices would soar, as would hourly incomes of real estate agents who are paid a percentage of the sales price. It would look as if real estate agents had become far more productive, when in fact they had not changed the quality of their ser­vices, so the apparent productivity growth would be a mirage. This sort of artificial increase in productivity is generally corrected when government statisticians calculate a price index for the real estate sector that takes this trend into account and adjusts the sector’s gross value added accordingly. Thus, price changes are fundamental to productivity growth, but any changes to quality need to be included too. If the ratio of quality to price increases, ­either b­ ecause prices fall or quality increases, t­ here is productivity growth. Another example from the auto industry is the not-­yet-­released Tesla Model 3, which is being advertised with a price of $35,000. Its features—­its quality—­are only slightly worse than the Model S, which costs $69,500.77 Thus, if e­ very consumer who would have purchased a Model S switched to

­Behind the Disconten • 31

the lower-­cost Model 3, the result would be a rise in living standards and economic growth. This has been happening with computer equipment. The cost of pro­cessors and other inputs has plummeted in recent de­cades, reducing the cost of computer devices relative to their computing power. Th ­ ose who buy computer devices are richer as a result. Productivity growth, therefore, can be mea­sured by growth in the ratio of quality to price.78 Inequity and Inefficiency

Both productivity measures—­g rowth in GDP per worker and changes in prices—­show declines in productivity growth in t­ hose sectors with the most top income earners. The inescapable conclusion is that top income earners are benefiting less from innovation than from inefficiency. Take value added per worker. As a starting point, consider that from 2000 to 2015, inflation-­adjusted GDP per worker grew slightly faster in the United States—22 ­percent over the entire period—­relative to the 15 original members of the Eu­ro­pean Union (EU 15), where it grew by 15 ­percent.79 In both regions, the growth was driven by advanced industries—­including ­those in manufacturing, computer ser­vices, information—­which are seeing fantastic productivity increases. In the United States, the information sector registered 151 ­percent productivity growth over the period. In the EU 15, it was 52 ­percent. Manufacturing saw 67 ­percent growth in the United States, and 33 ­percent growth in the EU 15. As we saw, roughly 20 ­percent of top income earners work in ­these sectors across the OECD countries and slightly less in the United States. Now, consider the sectors employing most top earners. GDP per worker in healthcare and social assistance increased by a meager 6 ­percent in the EU 15 and just 4  ­percent in the United States, far below aggregate productivity growth (figure 1.9). In both the EU 15 and the United States, the education sector registered negative productivity growth, −9 ­percent and −5 ­percent, respectively, and so did l­egal ser­vices (−3 ­percent and −13 ­percent, respectively). Computer ser­vices saw rapid growth, but all other professional ser­vices brought negative growth to the EU 15 (−4 ­percent) and modest growth to the United States (5 ­percent). Meanwhile, finance saw relatively strong productivity growth from 2000 to 2015, especially in the United States, but it is unlikely that this is the result of innovation in financial ser­vices; rather, it appears to have more to do with the rising return to capital incomes discussed above, as

Mining Miscellaneous professional, scientific, and technical services Real estate Computer systems design and related services Arts, entertainment, and recreation Finance and insurance Other services except government Information Professional services except computer Legal services Construction Transportation and warehousing Professional, scientific, and technical services Accommodation and food services Wholesale trade Government Educational services Administrative and support services Retail trade Health care and social assistance

U.S. productivity growth 2000–2015

0%

0%

16

14

0% 12

%

%

%

%

0% 10

80

60

40

20

0%

0% –2

–4

0%

Manufacturing

European Union (15 countries)

Figure 1.9. ​Growth in real GDP per worker by sector in 15 Eu­ro­pean countries and the United States, 2000–2015 Source: Eurostat, Bureau of Economic Analy­sis.

­Behind the Disconten • 33

­ able 1.4.  Sector-­specific price changes relative to total national price T changes, in selected OECD countries and the Euro Area

Education Health Housing ser ­vices Food Clothing and footwear Motor cars Communications

Germany

Italy

France

UK

Euro Area

United States

0.22 0.11 −0.01 0.02 −0.15 −0.11 −0.55

0.05 0.16 0.16 −0.01 −0.05 −0.09 −0.62

0.23 −0.02 0.14 0.02 −0.17 −0.15 −0.56

1.96 0.20 0.22 0.01 −0.65 −0.42 −0.36

0.20 0.08 0.07 0.01 −0.14 −0.16 −0.56

0.76 0.33 0.05 0.06 −0.39 −0.37 −0.25

Source: Eu­ro­pean Central Bank, Statistical Data Ware­house, May 1996 to May 2017 price index; values divide item growth rate by overall growth rate. U.S. data is from the Bureau of L ­ abor Statistics for the same period except Communications starts in 1998.

well as ­factors I discuss in chapter 9.80 Comparable data for Canada show the same pattern of slow productivity growth in the sectors of the one ­percent.81 Price data show the same pattern. Throughout the EU as well as the United States and the largest individual Eu­ro­pean economies, prices for education, health, and housing have exceeded total price growth. This means that t­ hese sectors are driving inflation and eroding the spending power of consumers and businesses. ­These sectors are becoming relatively less efficient compared with other sectors, such as clothing, motor vehicles, and communications. Clothing, motor vehicles, and communications ser­vices and equipment provide better value to consumers than they used to (­table 1.4). The price data suggest that the same c­ an’t be said for education, healthcare, and housing. The only serious objection to the conclusions offered above is that the price indexes ­don’t consider advances in quality. That is not a concern with housing, where quality is easy to mea­sure (by, for example, looking at price changes of the same unit over time), and statistical offices include quality changes. For education, ­there is no evidence to suggest that quality has improved in the United States and some grounds for concluding the opposite, as the administrative costs associated with higher and lower levels of education have grown sharply.82 It is pos­si­ble that t­ here have been modest gains in other OECD countries, but it is unclear if they would be enough to justify the price increases. On the surface, you might think it would be impossible to make education more efficient—to raise the quality relative to the price, but consider online low-­cost education ser­vices such as Coursera and Udacity, in

34 • Chapter 1

which highly qualified instructors provide clear technical lessons at a fraction of the traditional cost. Healthcare is more complicated. The productivity metrics described above arguably miss what r­ eally ­matters—­healthcare’s benefits to society. Indeed, the benefits of healthcare may be increasing as medical technologies and procedures improve. ­There is some evidence for this claim, particularly in Eu­ rope, where mortality rates, adjusted for age, continue to fall. Certainly, ­there have been many improvements in treating the diseases of old age through prescription drugs that lower blood pressure, for example, or surgical procedures. Infants are also considered ­viable at much ­earlier ages than they used to be. In the United States, however, the added benefits of healthcare have been very modest relative to the added costs. Mortality rates have not fallen for most age groups, only the el­derly and the very young. Subjective health status worsened between 1990 and 2015 for U.S. residents between the ages of 25 and 59, and the share of the population who report being unable to work as a result of disability increased from 4.4 ­percent to 6.8 ­percent between 1980 and 2015, ­after adjusting for age.83 Indeed, the best available evidence suggests that most new treatments and procedures introduced since 1976 are actually worse than the standard ones in terms of the ratio between quality and costs, according to a comprehensive analy­sis by the economist Tomas Philipson and his co-­authors Kristopher Hult and Sonia Jaffe.84 New procedures are usually slightly better than ­those they are replacing at extending life, but they typically cost several times more per quality-­adjusted life year, which is the standard metric for evaluating costs relative to benefits. In other words while patients may live slightly longer, most new treatments are less eco­nom­ically efficient. Thus, the most likely interpretation of the data described above is that healthcare and other domestic sectors have dragged down economic growth, even as they have been the principal source of income for the very rich. Economic growth and changes in the income distribution are extremely complex phenomena, which likely have many c­ auses. Some scholars, such as Robert Gordon, have argued that the productivity growth slowdown is a consequence of the fact the Industrial Revolution—­with its marvelous inventions—­can happen only once in history.85 Subsequent technological advances simply have not been as power­f ul. He may be right, and l­abor economists are likely correct that recent technological change has favored skilled workers, thereby exacerbating income in­equality.

­Behind the Disconten • 35

Still, t­ hese are at best incomplete explanations for slow growth and the extreme in­equality found in the United States and, to a lesser degree, other advanced countries. A more complete analy­sis must consider how po­liti­cal forces shape the economy, including the historic and con­temporary institutions that govern markets. As ­later chapters ­will discuss in detail, the institutions that govern the domestic ser­vice industries are controlled by them— a situation that allows elites to charge extraordinary and uncompetitive prices. Elite professions and some domestic industries operate as if they w ­ ere cartels. This allows individuals to get rich, while at the same time blocking innovations that would benefit society. As prices rise in ­these sectors, the purchasing power of consumers is slowly reduced and even undermined.

SLOWING PRODUCTIVITY GROWTH, RISING IN­E QUALITY, AND POISONED POLITICS

Trends within the United States illustrate how slow productivity growth combined with high and rapidly rising income in­equality can undermine demo­ cratic institutions. Confidence in elected officials, big business, and the media has plummeted since the 1970s, according to Gallup surveys, which have asked the same questions using the same basic methods for de­cades.86 The percentage of U.S. adults who are confident in Congress was over 40 ­percent in the early to mid-1970s (figure 1.10). A ­ fter a brief fall, it climbed back up to 41 ­percent in 1986, but since 1989, confidence in Congress has never risen above 30 ­percent, and in recent years it has hovered near just 10 ­percent. This is extraordinary considering that p­ eople responding to Gallup surveys are themselves responsible for selecting members of Congress. The trend reflects an increasingly polarized population. Similar declines in confidence have been observed for other impor­tant institutions, such as the press, large businesses, and even public schools. Confidence in the “medical system” has also plummeted from 80 ­percent in the 1970s to 36 ­percent ­today, as healthcare costs have soared. It would be tempting to dismiss this as a psychological change in attitudes to confidence or trust, but other institutions have remained popu­ lar. Confidence in the military has actually increased from 58 ­percent in the tumultuous 1970s to 74 ­percent in 2018, and confidence in the police has remained just above 50  ­percent since the early 1990s—­despite recent

36 • Chapter 1

Congress

Newspaper

2018

2016

2014

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1991

1989

1986

1984

1981

1977

1973

60 50 40 30 20 10 0

Big Business

Figure 1.10. ​Confidence in Congress, big business, and newspapers, among U.S. residents, 1973–2018 Source: Gallup, “Confidence in Institutions,” https://news.gallup.com/poll/1597 /confidence-institutions.aspx.

controversies. Small businesses are also widely popu­lar (confidence in 2018 is 67 ­percent) and more so now than in the early 1990s, perhaps ­because they are often engaged in intense competition with other businesses—­including large ones—­and often operate on relatively low profit margins. The fact that falling confidence in politics has coincided with rising in­ equality in the United States and other developed democracies does not mean that one has caused the other. Yet, I ­w ill pre­sent some facts that are consistent with the theory that in­equality has sapped confidence and, more importantly, undermined civic health. For instance, ­there is direct evidence that extreme income in­equality distorts politics by giving undue influence to the rich, who can readily outspend the vast majority of ordinary p­ eople. ­There are two reasons to be worried about this: The rich are more po­liti­cally engaged than the m ­ iddle and working classes, and they have dif­fer­ent policy preferences. ­People in the richest income group in the Current Population Survey w ­ ere 15 percentage points more likely to vote than ­those in lower groups, according to my analy­sis of the 2016 Voter Supplement.87 The situation is even worse, as the po­liti­cal scientists Benjamin Page, Larry Bartles, and Jason Seawright uncovered in a revealing study of the super-­wealthy. Not only do the top one ­percent vote at much higher percentages, they are also extremely engaged in making campaign contributions and coordinating such contributions, as well as contacting their elected officials and attending po­liti­cal events.88

­Behind the Disconten • 37

As for policy preferences, Page and his co-­authors find that the very rich are distinctly and almost unanimously concerned about the federal government’s national deficit—­which they may view as merely delaying ­future tax increases—­and less committed than the ­middle class to supporting the welfare state, including public spending on Social Security, healthcare, and education. It is in­ter­est­ing to note, though, that the majority of the super-­rich (62 ­percent) agrees that income differences in the United States are too large. The differences run deeper, as Gallup data shows by asking ­people to state the most impor­tant financial concern facing their f­ amily in their own words. Gallup has done this each year from 2006 to 2018 and received 33,000 responses. The data w ­ ere recoded into categories afterward by listening to the verbatim recordings. Stark differences emerge between the rich and the middle-­and lower-­ income classes. The rich, who are much less likely to have any financial prob­ lems, are overwhelmingly preoccupied with the ­future, while the ­middle and lower classes strug­g le with current expenditures on t­ hings like housing, healthcare, and paying off debts (­table 1.5). Among lower-­income families, 51 ­percent of their top financial concerns fell into ­these categories, but for the rich, t­ hese amounted to only 19 ­percent. ­These rates ­were almost reversed for ­future concerns, including retirement savings, tax obligations, and saving for college, which w ­ ere cited by 52 ­percent of rich families but only 31 ­percent of lower-­income families. ­Here the connection between lackluster productivity growth and po­liti­ cal concerns is clear. Rapid inflation in healthcare, starting around 1980, has resulted in skyrocketing prices, which are born by consumers, businesses, and the government. Healthcare went from 8 ­percent of GDP in 1976 to 18 ­percent 40 years ­later.89 What the U.S. dollar is spent on is very dif­f er­ent now compared with what it was spent on in the past. Spending has shifted away from cars and appliances, food and clothing and t­ oward insurance, healthcare, education, and shelter. ­These changes are felt by the m ­ iddle classes, most of whom rate their personal financial situation rather poorly when asked by Gallup. Economic in­equality played a complex role in the 2016 U.S. presidential election. Republican Party primary voters rejected the proposals supported by the party’s upper-middle-­class leadership—­such as reducing the bud­get deficit, expanding trade agreements, and making reforms to immigration that would increase ­legal channels to work and provide a path to citizenship for

38 • Chapter 1

­ able 1.5.  The financial health and most impor­tant financial prob­lems T facing dif­fer­ent income groups in the United States, 2006–2018

All

High income (income of $250,000 or higher)

­Middle to upper-­middle income ($50,000 to $249,000)

Low income (below $50,000)

Personal financial situation is good or excellent

47%

84%

65%

28%

No financial prob­lems

15%

25%

15%

12%

Prob­lems meeting current expenses Cumulative current expenses

43%

19%

36%

51%

Healthcare Lack of money Debt Housing High cost of living

12% 9% 9% 7% 5%

7% 2% 4% 4% 3%

11% 5% 8% 7% 5%

13% 13% 11% 8% 6%

Prob­lems meeting ­f uture expenses Cumulative ­f uture expenses

36%

52%

40%

31%

Retirement, lack of savings, interest rates, state of the economy, or stock market Saving for college Taxes

26%

30%

26%

26%

6% 3%

12% 10%

10% 4%

3% 2%

Source: Gallup Poll Social Series, 2006–2018. The items are “How would you rate your financial situation ­today–as excellent, good, only fair, or poor?” and “What is the most impor­tant financial prob­lem facing your ­family ­today?” Open-­ended responses w ­ ere subsequently recoded by Gallup into vari­ous categories. Sample size is 33,464 respondents interviewed by phone via random-­digit dialing.

the nation’s undocumented residents. Instead, the party’s middle-­and lower-­ income base embraced Trump’s aggressively nationalist rhe­toric about reducing the flow of immigrants by building a wall on the Mexican border, deporting them, and enacting tariffs on imports. They also strongly supported Trump’s proposals to eliminate the existing healthcare system and to invest heavi­ly in both the military and in public infrastructure. Republicans with modest and low incomes w ­ ere much more likely to tell Gallup that ­these Trump proposals w ­ ere very impor­tant to them, relative to the Republicans with over $250,000 in h ­ ouse­hold income.90

­Behind the Disconten • 39

Trump was the only candidate except Ted Cruz—­the other Republican Party front-­runner—to see larger support in low-­and middle-­income ­house­holds than in rich ­house­holds among core party supporters. For ­every other candidate, rich members of his or her party w ­ ere more favorable about the candidate than less affluent party members (figure 1.11). As for the general election, Trump won largely b­ ecause of success with middle-­income voters. I analyzed candidate favorability ratings from Gallup surveys conducted between October and November 2016 and weighted the approximately 33,000 responses by the probability of voting in the general election. Hillary Clinton had much higher favorability ratings from rich voters ($250,000 and above) and low-­income voters (less than $36,000). Her margins w ­ ere 10 and 6 percentage points, respectively, for t­ hose groups, but she tied Trump votes in the ­middle of the income distribution. Moreover, in swing states, Trump had a three-­percentage point advantage among middle-­ income voters, even as Clinton won narrowly among the rich and tied Trump among low-­income swing state voters.

THE PO­L ITI­C AL CONSEQUENCES OF RACIAL IN­E QUALITY

I believe the multi-­decade slowdown in economic growth and the related rise in income in­equality spurred mass po­liti­cal dissatisfaction, while draining confidence and trust in national governments, the media, and other once-­ respected institutions, as conventional po­liti­cal parties, party leaders, and thought leaders failed to devise and implement solutions that would enhance economic growth and allow the gains to be more broadly shared across the population. Yet, dissatisfaction with conventional po­liti­cal platforms need not have led to a surge in far-­right nationalist party popularity—as it has in a number of Eu­ro­pean countries—or allowed a nationalist Republican Party candidate like Donald Trump to win his party’s nomination and the general election. The deeper po­liti­cal context of votes for the far Right in Eu­rope and for Donald Trump in the United States is that Eu­ro­pean p­ eople and their descendants comprise the demographic majority in most of the world’s rich countries and all but a handful of the world’s rich democracies. Thus, when citizens in European-­majority democracies consider policies such as openness to trade and immigration, minority rights, or access to markets and benefits,

56.7%

67.8%

Carson

70.7%

72.7%

88.7% 82.5%

Sanders Clinton Democrats Low income

Trump

64.1%

72.4%

Figure 1.11. ​Favorability of 2016 presidential candidates in their own party by income group

Middle to high income

Kasich Rubio Republicans

Very high income

60.1%

Cruz

57.1%

67.0%

77.8%

82.3%

Source: 2016 Gallup Daily Tracker. Sample size is approximately 113,000 for Clinton and Trump favorability scores and varies for the other candidates relative to their perceived viability at vari­ous points in the primary election. Income is defined for the ­house­hold as low if u­ nder $36,000 and very high if in the highest category of $240,000 or more. Every­one ­else is in the middle- to high-­income group. The favorability item response was weighted using both the Gallup weight and in the reported data above using the probability of voting in the 2016 election. To calculate this, I matched data from the Current Population Survey’s 2016 voter supplement to Gallup data using race and Hispanic status, educational attainment, age in ten-year bands (­a fter 18–25), and income in three groups that roughly correspond to the groups above.

Bush

72.6%

80.3%

85.8%

­Behind the Disconten • 41

it is generally from a perspective of relative power and privilege compared with ­people from other groups. For a subset of ­people of Eu­ro­pean ancestry, many of whom are struggling to pay their bills, the thought of p­ eople from poor countries moving into theirs provokes fear of the economic consequences, including w ­ hether their governments can continue to meet the obligations of the welfare state, or, in the United States, Medicare and Social Security. ­There is also strong evidence that issues tied to ethnic differences and attitudes are impor­tant in shaping po­liti­cal preferences for nationalism and are largely in­de­pen­dent of economic concerns. H ­ ere, I w ­ ill quickly review the evidence. First, ­there is virtually no relationship between the probability of voting for Trump or a far-­right candidate in Eu­rope and working in the one sector that has been the source of the most upheaval in Eu­rope and the United States: manufacturing.91 Working in manufacturing is among the least predictive ­factors of voting preferences mea­sured in large-­scale surveys. To emphasize the point further, consider that only 3.5 ­percent of American adults who hold a favorable view of Trump report to Gallup that they work in manufacturing or production compared with 2.8 ­percent of t­ hose who hold an unfavorable view. Moreover, living in regions of the country that have experienced larger job losses in manufacturing or increased import competition from China does not predict greater support for Trump, as I found in a large-­scale analy­sis of 125,000 U.S. adults using Gallup data.92 Geography is likewise often seen as an impor­tant ­factor, but county population density is weakly predictive of lower support for Trump. Second, the relationship between income and nationalist support does not suggest deprivation or poverty is a strong motivating ­factor for most supporters of nationalist or far-­right parties, but t­ hose voters are more likely than moderate voters to have missed out on the gains from economic growth. In Eu­rope, 6.5 ­percent of far-­right voters are in the top 10th percentile of the national income distribution, according to my analy­sis of the Eu­ro­pean Social Survey, which suggests that they are much less likely to have benefited from rising income in­equality than other voters.93 Likewise, only 27 ­percent report living comfortably on pre­sent income (versus 33 ­percent of non-­far-­right voters). Yet, controlling for education and age removes any significant relationship between voting preferences and t­ hese variables. Likewise for Trump, subjective assessments of personal financial conditions ­don’t explain support for him. Republicans—­and Trump supporters

42 • Chapter 1

in particular—­rated their personal financial situation very poorly while Obama was president, but that assessment r­ ose immediately ­after Trump won and remained high thereafter, despite no a­ ctual change in prosperity relative to Demo­cratic Party supporters. As the Gallup Poll’s former Editor in Chief Frank Newport put it, “[Americans] view their own personal finances through the lens of partisanship, becoming more positive when the president is from their party and more negative when he is not.”94 Using more objective data, I find that personal income has a modestly positive effect on the probability of holding a favorable view of Trump, mostly ­because Trump supporters are much less likely to have low incomes than ­those who ­don’t support him. Importantly, however, they are less likely to have very high incomes ($240,000 or above) and tend to live in zip codes with lower average incomes, as reported by the Internal Revenue Ser­vice. Fi­nally, objective mea­sures related to racial isolation and subjective responses to survey items about race are impor­tant predictors of nationalist support and much more impor­tant than income. One of the variables that most strongly predicts support for Trump overall among white p­ eople and even for white Republicans is the share of white p­ eople living in the neighborhood of the respondent. For e­ very standard deviation increase in the zip code white population share (25 ­percent), the probability of favoring Trump increases by 12 percentage points among non-­Hispanic white respondents. This effect controls for ­every observable demographic characteristic and several other geographic ones.95 In Eu­rope, ­those who voted for right-­wing party candidates in the last election are 14 percentage points more likely to say that Eu­ro­pean ancestry is an impor­tant criterion for immigration policy compared with ­those who voted for other parties. They are also significantly more likely to say that races differ by work ethic and intelligence.96 Many U.S. adults of Eu­ro­pean ancestry who identify with the Republican Party—­which contains both ­those on the moderate and far right on the po­liti­cal spectrum—­report that black ­people are lazier than other groups (42 ­percent) or lack the ­will power to move out of poverty (55 ­percent).97 A substantial fraction rate black ­people as less intelligent than white ­people (26 ­percent).98 Republicans who supported the candidacy of Donald Trump w ­ ere even more likely than o­ thers in their party to express racial bias or lack of sympathy t­ oward black p­ eople. For example, white Americans who reported a

­Behind the Disconten • 43

favorable opinion of Donald Trump during his campaign for president ­were 13 percentage points more likely to say that black c­ hildren have the same opportunities to receive a good education as white ­children compared with other white ­people affiliated with the Republican Party, and they ­were 37 percentage points more likely to hold this view compared with white p­ eople who supported the Demo­cratic Party.99 An excellent book on the U.S. 2016 presidential campaign by John Sides, Michael Tesler, and Lynn Vavreck assem­ bles still more compelling evidence along ­these lines across a large number of surveys and sources.100 For further evidence that racial and cultural differences are impor­tant to nationalist politics generally, consider the ­actual party platforms, speeches, and manifestos of nationalist leaders. Researchers at the Wissenschaftszentrum Berlin für Sozialforschung (WZB, or Berlin Social Science Center) created Proj­ect Manifesto, which codes the public documents and speeches of po­liti­cal parties from around the world. The parties they classify as “nationalist,” based on ­these materials, received 9 ­percent of votes in the most recent elections in 19 OECD countries. The database breaks down po­liti­cal documents by sentence-­like fragments and uses country-­level experts to code t­ hese statements by theme. I used ­these data to compare the content of nationalist party materials with ­those of other parities. Strikingly, I found that nationalist parties in the OECD countries are 12.6 times more likely than nonnationalist parties to make statements that immigration is harmful to the national way of life. Such statements comprised approximately 4 ­percent of all po­liti­cal sentences among nationalist parties (figure 1.12). Other distinctive themes raised by nationalist party materials include a high percentage of statements against multiculturalism and internationalism, as well as statements in ­favor of protectionism (and so against trade); the nationalist parties w ­ ere far less likely than other parties to make statements favorable to minority groups or statements in ­favor of welfare, equality, or education. ­There can be no doubt that views regarding cultural, ethnic, and racial differences have a large effect on policy attitudes. Not many generations have passed since Eu­ro­pean countries occupied much of the world and exerted nondemo­cratic influence over many non-­European p­ eople. Likewise, many African Americans living ­today remember the overt formal and informal prejudice of the Jim Crow era, which waned in the 1960s but lingers in vari­ous forms ­today.

Immigration harmful to national way of life Against multiculturalism Favor national way of life Against internationalism Favor protectionism Favor traditional morality Middle-class and professional groups Anti welfare Free market Anti education Culture Tech and infrastructure Demographic groups Pro education Favor welfare Pro equality Favorable toward minority groups Favor multiculturalism Against protectionism Against traditional morality Critical of national way of life 0

2

4

6

8

10

12

14

Share of quasi-sentences on subject for nationalist parties/share for non-nationalist parties Figure 1.12. ​Share of OECD nationalist party public speeches and documents devoted to vari­ous positions relative to share from all OECD parties, 2000–2017 Source: Data from Andrea Volkens et al., “Manifesto Project Main Dataset (Party Preferences),” Manifesto Project Dataset (version 2018b) (Berlin: Wissenschaftszentrum, 2018), https://doi.org/10.25522/manifesto.mpds.2018b.

­Behind the Disconten • 45

The in­equality in income and social status analyzed in this chapter is inextricably tied to racial and ethnic differences both across and within countries. Across countries, ­there are massive gaps between the richest regions of the planet—­North Amer­i­ca, East Asia, and Northwestern Europe—­and the rest, and within countries ­there remain large gaps in social status between ­those of Eu­ro­pean or East Asian descent and other groups. Brown University economists Louis Putterman and David Weil have constructed mea­sures of racial diversity across global countries and find that racial diversity strongly predicts income in­equality and that p­ eople with ancestry from poorer regions of the world tend to occupy lower social status in the countries where they now live.101 That pattern is evident in Eu­rope, using data from the Eu­ro­pean Social Survey (­table 1.6), and with some exceptions, to be discussed l­ater, it holds for the United States, where African Americans and Hispanic Americans generally have lower incomes and levels of education than Eu­ro­pean Americans and East and South Asians. In Eu­rope, ­people with ancestry from a Sub-­Saharan African country or region are much less likely to be in the top 10th percentile of the earnings distribution relative to t­ hose of West Eu­ro­ pean or Northern Eu­ro­pean ancestry. In the United States, the United Kingdom, and Eu­rope, ­people descended from West African slaves—­who w ­ ere forcibly brought to British colonies and the Amer­i­cas—­still suffer from discriminatory attitudes that affect public policy, the provision of public goods, and access to markets, as l­ ater chapters ­will discuss Fully 81 ­percent of African Americans living in the United States say that racism is widespread, according to Gallup survey data.102 Discrimination appears to be less overt in Eu­rope, but p­ eople of African descent are disproportionately victims of it, according to data from the Eu­ro­pean Social Survey.103 Eu­ro­pean residents descended from Sub-­Saharan Africans—­and especially ­those who ­were likely to have been enslaved—­are far more likely to report being victims of discrimination based on their race, ethnicity, or nationality (figure 1.13). Roughly half (54 ­percent) of black p­ eople from the United States, Canada, and the Caribbean—as well as ­those of general West African or Sub-­Saharan regional ancestry—­report being discriminated against in their country of residence. The figure is high but much lower for Sub-­Saharan Africans who name a specific country as their ancestral origin (30 ­percent), lower still for East and South Asians (24 ­percent and 21 ­percent, respectively) and other groups, and negligible for ­those of Eu­ro­pean ancestry.

46 • Chapter 1

­ able 1.6.  House­hold income status and immigration T status of Eu­ro­pean residents by ancestral region, 2016

East Asian North Eu­ro­pean West Eu­ro­pean East Eu­ro­pean South Asian South Eu­ro­pean Likely African slave ancestry Latin American Sub-­Saharan African West Asian or North African

­ ercent in top P 10th percentile of national income distribution

Born in country of residence

13 13 10 9 8 7 7 5 5 5

12 96 95 92 32 89 53 25 17 48

Source: Eu­ro­pean Social Survey 7. Sample consists of 20 Eu­ro­pean countries plus Israel. African slave ancestry determined for respondents who say their ancestry is “African American,” “Black Canadian,” or from a predominantly black Ca­rib­bean country (e.g., Bahamas, Barbados, Jamaica, Trinidad, Haiti), or from the region of Sub-­Saharan Africa or West Africa. The post-­stratification weight was combined with the population size weight to produce ­these estimates. Estimates shown only for groups with at least 50 observations.

This is not to suggest that discrimination applies only to ­those with Sub-­ Saharan African ancestry. Data from the Pew Research Center show that other groups confront unfavorable views across Eu­rope.104 ­These include the Roma, who are thought to have originated in India, as well as Muslims and Jews, both of whom have ancestral or at least cultural origins largely from North Africa and West Asia. When Eu­ro­pe­ans ­were asked by Pew about what immigrants want in 2014, the majority of Italians, Germans, and French selected “be distinct from our society” rather than “adopt our customs and way of life.” In the UK, 47 ­percent chose “distinct” and 34 ­percent chose “adopt.” 105 Detailed evidence from Italy shows real-­world consequences of t­ hese attitudes. Harvard economist Alberto Alesina and his collaborators gave implicit association tests to school teachers in Northern Italy, which are meant to mea­sure partially unconscious automatic responses to stimuli. They found that 67 ­percent of teachers exhibited severe bias against immigrants, meaning they w ­ ere highly likely to associate immigrants with bad t­ hings. Teachers with greater implicit bias also gave out lower math grades, even ­a fter

­Behind the Disconten • 47

60%

54%

50% 40% 30%

30%

24%

20%

21%

19%

17% 2%

2%

2%

1%

West European

East European

North European (Nordic)

Latin American

West Asian or North African

South Asian

East Asian

Sub-Saharan African

Likely African slave ancestry

0%

South European

10%

Figure 1.13. ​Percentage of population living in Eu­rope or Israel who believe they are discriminated against based on their race, ethnicity, or nationality, 2014 Source: Eu­ro­pean Social Survey 7. Discrimination is based on the following question: “Would you describe yourself as being a member of a group that is discriminated against in this country?” The respondent must answer yes and indicate that the basis for this discrimination is their nationality, ethnicity, race, or color.

controlling for test scores on standardized math exams.106 In a separate study with Stephanie Stancheva and Armando Miano, Alesina surveyed adults in the United States, the United Kingdom, Sweden, Germany, Italy, and France, asking them to estimate the characteristics of immigrants, including their education, employment status, reliance on welfare, and percentage who are Muslim and Christian. In ­every country, native residents tended to exaggerate the economic weaknesses of immigrants and their religious differences from the white population. Right-­wing respondents w ­ ere especially likely to make ­these ­mistakes, and they made larger ­mistakes than other respondents, suggesting biased po­liti­cal attitudes distort their perceptions of real­ity at the expense of immigrants or potential immigrants.107 Along ­these lines, racial chauvinism—if not outright racism—is closely linked to right-­wing politics both in the United States and in Eu­rope. Transparent support for notions of white superiority and nationalism can be found among a subset of Trump supporters who identify with a far-­right movement known as the “Alt-­R ight.” The movement was described and

48 • Chapter 1

summarized in the far-­right publication Breitbart, which is owned and developed by one of Trump’s closest po­liti­cal advisors, Steve Bannon, who is no longer working in the Trump administration.108 Though ­there is diversity in its approaches and influences, the following are among the princi­ples associated with the alt-­right movement: (1) Racial and ethnic hierarchies are natu­ral and related to ge­ne­tic diversity; (2) it is far more natu­ral to care about your own demographic in-­group, rather than a h ­ uman being as such; (3) ­free markets should be subordinate to in-­group cultural concerns and preferences for ethnic isolation; (4) race determines culture, and thus white ­people who want to preserve their culture should not mix with or integrate with other races. According to a 2017 online survey conducted by psychologists P.  S. Forscher and N. Kteily, who collected data from approximately 500 ­people who identified as Alt-­R ight, 89 ­percent ­were white and roughly 63 ­percent ­were male.109 Among ­those who voted in 2016, 84 ­percent of the Alt-­R ight voted for Donald Trump; 40 ­percent rated “ge­ne­tic inferiority” as a trait that better describes black ­people than white ­people on a scale of 1 to 7; and 60 ­percent rated black ­people as less “rational and logical” than white ­people. When asked to rate groups on a scale of 0 to 100, with 0 being something like a chimpanzee and 100 being a modern Homo sapiens, alt-­right supporters expressed a clear racial hierarchy, with white ­people and Eu­ro­pe­ ans at the top and Muslims, Arabs, Nigerians, black p­ eople, Mexicans, and Jews at the bottom (figure 1.14). Among alt-­right supporters, 64 ­percent rated white ­people as more evolved than black ­people. ­Those results can be compared with a separate survey of white Americans conducted by Ashley Jardina of Duke University and Spencer Piston of Boston University. They found that 52 ­percent of white Trump supporters rate black ­people as less evolved than white ­people. This share was smaller but still very high among white Republicans (39 ­percent) and white Demo­crats (33 ­percent). Jardina and Piston, writing with Sean McElwee, conclude that dehumanizing views are “pervasive across white social groups.” In open-­ended responses, ­those who indicated that white ­people ranked higher mentioned such notions as black ­people lack intelligence or morals, according to the authors.110 Data from the previous de­cade show a similar and substantial number of white p­ eople harbored racist views of some kind. When asked in 2003 and 2004 why black ­people have lower incomes and worse housing than white

Christians

87

87

90

Whites

Swedes

86

Americans

82

Men

82

Europeans

82

Women

Mexicans

78

Russians

67

72

Jews

65

Blacks

59

Nigerians

Muslims

56

59

Arabs

­Behind the Disconten • 49

Figure 1.14. ​How the average alt-­right supporter ranks h ­ uman groups on 0–100 scale where 0 is a chimpanzee and 100 is a human Source: Forscher, P. S., and N. Kteily. “A Psychological Profile of the Alt-­R ight.” August 10, 2017. Retrieved from osf.io/preprints.

p­ eople, roughly one-­third (38 ­percent) of white respondents agreed that “fundamental ge­ne­tic differences between the races” explained at least a ­little of the gap, according to survey data collected by po­liti­cal scientists Leonie Huddy and Stanley Feldman. A similar share (35 ­percent) attributed at least a ­little of the black-­white gaps in test scores to fundamental differences, with 2.9 ­percent saying fundamental differences explain “a g­ reat deal” of the gap and 17.5 percent choosing “some.”111 The same authors report evidence that ­these overtly racist attitudes predicted opposition to policies that would increase economic opportunities for black ­people—­such as integrated housing and more spending on schools. Interestingly, most alt-­right adherents ­don’t consider themselves racist, ­because they claim to avoid discrimination in their personal lives, even as they hold views that are the textbook definition of racism. They thus distinguish themselves from a neo-­Nazi subset of the Alt-­R ight, who are actively hostile ­toward t­ hose who are not white. The Forscher and Kteily survey suggests that principled nonprejudice in individual interactions with black ­people is not as impor­tant as the article implies, although this attitude is common among roughly half of t­ hose in the Alt-­R ight. Just ­under half of alt-­right supporters (45 ­percent) expressed some level of agreement with the statement “Being

50 • Chapter 1

non-­prejudiced t­ oward Black p­ eople is impor­tant to my self-­concept,” even as 38 ­percent rejected it, and the rest gave a neutral response (17 ­percent).112 ­W hether ­these responses are described as racist, ethnocentric, or other­ wise, they have impor­tant policy implications with re­spect to immigration, education and training programs, and welfare. White Trump supporters are 11 percentage points more likely than other white Republicans and 27 percentage points more likely than white Demo­crats to say that the government has “no role” to play “in trying to improve the social and economic position of blacks and other minority groups in this country.”113 In short, ­after de­cades of slow growth and rising in­equality, and a recent and large financial recession, nationalist voters in the United States and Eu­ rope are broadly unhappy about their financial conditions and have benefited less from economic growth than rich voters but are fairly typical eco­nom­ically. More distinctively, issues tied to race, culture, and immigration loom large in explaining nationalist attitudes. In this way, nationalist voters tend to be victims of elite-­driven in­equality and beneficiaries and defenders of race and nation-­based in­equality. Politics is likely to remain unhealthy ­unless both types of inequalities are combatted.

2 The Natu­ral Foundations of a Just Society

Reflecting on the realities of Jim Crow Amer­i­ca and the psychological tension it created, the author James Baldwin wrote: “Injustice is a commonplace.” And yet, “one must never, in one’s own life, accept . . . ​injustices as commonplace.”1 Amid the po­liti­cal turmoil gripping con­temporary democracies, along with stagnant economic conditions, rising in­equality, and dissatisfaction with government institutions, I believe ­there is a need to move beyond superficial policy discussions and revisit what it is we hope to achieve through our lives and po­liti­cal institutions and how to do so justly. As I develop l­ ater, the prob­ lems outlined in the previous chapter are, in large part, grounded in commonplace injustices that are repairable. In both ancient and modern writing, justice is regarded as a principal virtue of individuals and po­liti­cal entities. Yet, in this age of irony, skepticism, and post-­modern sensibility, it seems embarrassingly antiquarian to even pretend it exists. But, at the risk of embarrassment, I say it does. A sense of justice can be found in ­every corner of the world in ­every era of history. It can even be found in our ancestral DNA and in our neurons. Justice has a material real­ity like health. Much as ill-­health weakens the body, deprives it of normal functioning, and exposes its vulnerabilities, so injustice weakens social ties, dissuades ­people from investing in the ­future, prevents its victims from optimal functioning, and endangers groups of ­people. Even if you believe my interpretation of justice goes too far, I hope you ­will at least agree that a robust sense of justice is found across ­human populations. ­Whether justice be thought of as natu­ral or metaphysical, or as an emergent concept dependent on specific historical circumstances, it is rooted

52 • Chapter 2

in princi­ples and convictions that are widely shared across humanity and fundamental to the basic organ­ization of a healthy society. It is this sense of justice that I believe leads to an egalitarian but not strictly equal economy.

UNIVERSAL MORALITY

If justice is an inherent quality of ­human life, as I maintain, an understanding or sensitivity to justice should be ubiquitous across cultures and socie­ties, even as the unique history and circumstances of a society shape views about how justice should be understood and implemented. Aristotle wrote extensively about justice, and his work offers one of the earliest and lengthiest meditations on the subject for which written rec­ords exist. Modern scholars find two key aspects of justice in his work: One concerns conformity with laws that serve the common good; the other, proportionality or fairness, especially with re­spect to the rewards of merit.2 The King James translation of Jewish religious texts provides clear illustrations of justice as conformity with law and proportionality in market exchanges.3 Reciprocity in this broader sense, and in the narrower sense of exchange and trade of valuable commodities or ser­vices, has been deemed a h ­ uman universal. Reciprocating behavioral norms are found across ­every society studied by anthropologists.4 A related ethical norm is the “golden rule,” which implies some mea­sure of selflessness, in that it stipulates that one should treat ­others as one would oneself. This is more advanced than straightforward reciprocal exchanges of ­things that are of equal value and requires an ability to empathize with ­others. The golden rule is a well-­known teaching of Jesus of Nazareth, who urged his followers to “love thy neighbor as thyself,” but variants of the rule have been found in almost ­every religious and social tradition, such that it too is likely a universal aspect of ­human culture.5 Well-­documented examples of the golden rule have been found in the writings of the ancient Chinese phi­los­o­ pher Confucius, ancient Syrian, Persian, Indian (Hindu and Buddhist), and Greek texts, as well as in the oral tradition of West Africa.6 Since it is prominent in Chris­tian­ity, it should not be surprising that both Jewish and Muslim texts include the idea. The golden rule also emerged in secular republican Rome, in­de­pen­dently of Judeo-­Christianity. The classical Roman senator, politician, and phi­los­o­

Natu­ral Foundations of a Just Society  •  53

pher Cicero developed his own reciprocal ethical aphorisms. In an essay known as “On Duties,” which was nominally a letter addressed to his son in 44 BCE, Cicero wrote, “No one do harm to another, u­ nless provoked by injury,” “­Those who have the same virtuous desires and purposes love one another as they love themselves,” and “­There is no more essential duty than that of returning kindness received.”7 Likewise, he told his son, “One must not injure anyone e­ lse for one’s own profit,” adding that this idea was a natu­ral law with international validity.8 Further, Cicero emphasized a nonreciprocal ele­ment of magnanimity, also found in many religious traditions, which is to treat the poor with generosity: “But alike in bestowing benefit and in returning kindness, other ­things being equal, it is the highest degree incumbent upon us to do the most for ­those who need the most.”9 Moreover, he wrote, “Justice is to be maintained even t­ oward ­those of the lowest condition.” Mentioning slaves in par­tic­u­lar, he said that they o­ ught to be paid for their l­abor and assigned daily tasks in advance as if they w ­ ere “hired servants.”10 ­These statements together point to a core ele­ment of the golden rule that distinguishes it from fair exchange: empathy for the plight of the weak. Though Cicero wrote just before the birth of Christ, this is similar to the Catholic ethical teaching known as the preference option for the poor, which instructs Catholics to give greater weight to the preferences of the poor.11 The option for the poor can be thought of as rebalancing the scales to their benefit, especially in the marketplace, where their poverty is a clear constraint. It is quite likely to have been widespread throughout ­human cultures, as Cicero claimed, but b­ ecause literacy was rare u­ ntil recently, it is difficult to establish documentary evidence. Still, we know it has deep ancient roots. An inscription attributed to the Egyptian king Amenemhat I, who died roughly 2,000 years before Christ (in 1976 BCE), boasted: “I gave to the destitute, and raised up the orphan. I promoted the man with nothing as much as the man of means.”12

NATU­R AL JUSTICE

The ubiquity of the golden rule and related precepts suggests that this notion of justice is an impor­tant and universal characteristic of h ­ uman existence. Consequently, altruism should be a natu­ral aspect of ­human moral perception,

54 • Chapter 2

as Cicero thought. That is, perhaps h ­ uman minds grasp gradations of justice along a spectrum of reciprocating be­hav­ior, just as the h ­ uman eye naturally distinguishes color along gradations of the electromagnetic spectrum. If so, we could expect to observe the recognition of something like fairness among nonhuman animals and across all h ­ uman socie­ties. In a remarkable experiment, primatologists Sarah Brosnan and Franz de Waal trained female capuchin monkeys to exchange tokens for cucumbers, which they readily did in 95 ­percent of cases.13 Brosnan and de Waal then changed the price of the token for the capuchin’s neighbor but not for her. Specifically, they allowed the neighbor to receive a grape—­which ­these monkeys prefer over cucumbers—in exchange for the token. Video evidence presented by de Waal shows that this new deal did not go down very well.14 The capuchin stuck with the bad deal became visibly upset upon receiving the cucumber instead of the grape. Despite having enjoyed the cucumber seconds ago, she now flung it out of the cage. Her gestures made it clear that she wanted the grape instead. When made the same offer, she did the same t­ hing again, strenuously rejecting the trade. This rejection of the trade was observed far more frequently when the payout for the token was lower for the subject. Brosnan and de Waal interpret this experiment as evidence that the monkey recognized when it was being treated unfairly or at least not in the same way as the other monkeys. By switching the payoff from cucumbers to grapes, the experiment increased the value of the tokens, which is analogous to economic growth, but the gains of growth had gone only to the monkeys that received grapes and that did nothing to deserve t­ hose gains relative to t­ hose in the other group. Despite no material reduction in their living standards, the monkeys that received only cucumbers ­were less happy. Primates also exhibit be­hav­iors that are closer to the golden rule—­namely, generosity ­toward o­ thers. In one experiment, capuchins ­were given a choice between one of two tokens. One of the tokens resulted in a reward for that monkey but not the monkey next to it, vis­i­ble but in another cage. The other token resulted in both monkeys being rewarded. Thus, ­there was a 50–50 chance of prosocial be­hav­ior and selfish be­hav­ior. In the course of 30 rounds of this game, the monkey chose prosocial be­hav­ior much more frequently than random chance would have allowed and nearly 70 ­percent of the time by the last ten t­ rials. The greatest amount of sharing was with kin, followed by non-­ kin monkeys within the same group. ­There was no prosocial be­hav­ior with

Natu­ral Foundations of a Just Society  •  55

strangers. Likewise, when the rewards w ­ ere unequal, prosocial be­hav­ior also declined to the point that it could be attributed to random chance. The results imply that capuchins have preferences to bestow benefits to t­ hose in their group but not outsiders, provided the rewards are equitable.15 In another experiment, marmoset monkeys displayed even more generosity. They ­were frequently willing to pull a tray that gave food to a partner but nothing for themselves, and they w ­ ere far more likely to do this when it actually benefited another monkey than in the control condition in which the adjacent cage was empty and no monkey benefited.16 Thus, ­these monkeys displayed a willingness to contribute to the welfare of o­ thers, even when d­ oing so resulted in no immediate benefit to themselves. The extent to which primates recognize and respond to fairness is unresolved in the scientific lit­er­a­ture, but t­ here is clear evidence of some recognition. In addition, be­hav­ior that helps o­ thers at one’s own expense is commonly observed.17

­H UMAN FAIRNESS IN EXPERIMENTAL AND CULTURAL SETTINGS

­Humans also show a clear and consistent awareness of unfairness. An innovative experiment conducted by the Swiss economists Ernst Fehr and Simon Gächter recruited 112 Swiss students to play a game for two hours and earn a modest amount of money.18 In the game, four players w ­ ere given 20 francs each. They ­were told to decide how much they wanted to contribute to the group, knowing they would get only 40  ­percent back from what they contributed, while every­one ­else in the group would also get 40 ­percent but without giving anything up. In this sense, the group contribution is akin to taxation in support of a “public good,” such as infrastructure, security, and education, but it is valuable only if many ­people chip in, and in this setup, ­doing so is voluntary. This game was played ten times in a row, and the players could keep their money at the end. ­Table 2.1 shows the hy­po­thet­i­cal payoffs for individuals who chose to be completely selfish or generous ­under the extreme scenarios in which the group was ­either entirely selfish or generous. It’s clear that average income was highest when every­one donated (32), while the individual best-­case-­scenario (44) involved that person donating nothing, even though that harmed the group. Thus, for ­those motivated by self-­interest alone t­ here are strong incentives to donate nothing.

56 • Chapter 2

­Table 2.1.  Hy­po­thet­i­cal income of public goods game for extreme cases Every­one ­else donates all their money

No one ­else donates

32 44

0 20

32 11

17 20

Your income Donate all your money Donate nothing

Average group income Donate all your money Donate nothing

Source: Based on par­ameters from Fehr, Ernst, and Simon Gächter. “Altruistic Punishment in H ­ umans.” Nature 415.6868 (2002): 137–140; Fehr, Ernst, and Simon Gächter. “Cooperation and Punishment in Public Goods Experiments.” The American Economic Review 90.4 (2000): 980–994.

­ eople who started the game acting generously learned the strength of P ­these incentives and quickly ­stopped donating. The average contribution in the first round was just u­ nder half of total wealth (nine francs) but fell to around 10 ­percent of total wealth (two francs) by the tenth round. In modern states, as well as less formally structured tribal socie­ties, humans have mechanisms that range from criticism and shaming to ostracism, fines, prison time, and even execution to force compliance to support public goods and induce sharing.19 So, Fehr and Gächter tested an alternative game that was exactly like the first but allowed the players to punish one another. A donor in the first round could decide to punish a nondonor by taking away three of his or her francs, but as in real life, punishment was itself costly: a punisher had to forgo one franc. In contrast to a standard economic theory that ­people are selfish, many ­people ­were willing to punish nondonors. The results ­were spectacular, and it did not ­matter which version of the game ­people played first. The average donations ­rose from just over 50 ­percent in the first round to nearly 100 ­percent by the final round. In short, p­ eople w ­ ere willing to pool almost all of their resources when they ­were assured of two ­things: (1) ­Doing so would make every­one better off, including themselves, and (2) cheaters ­wouldn’t get away with it. Follow-up research by the same authors found that cheating elicited strong expressions of anger in the subjects, which may have contributed to

Natu­ral Foundations of a Just Society  •  57

their willingness to altruistically punish. (I draw on the scientific lit­er­a­ture’s definition of altruism as be­hav­ior that benefits ­others, such as the group, while imposing a cost on the helper).20 This and similar experiments have been tested many times and across many cultures, including in relatively poor and nonindustrialized socie­ties.21 Consistent evidence emerges that ­humans reject unequal distributions when they are not based on merit and that they cooperate to enhance the strength of their group and help ­others in need.22 In experimental settings, seeing another person in pain elicited a distinctive blood-­flow pattern in the brain, which researchers distinguished from cases in which the subject’s own pain was taken away by generous be­hav­ior.23 Both reciprocating gratitude and empathetic generosity ­were correlated with similar neural activity across individuals. Further evidence that ­there is a biological basis for altruism comes from studies of extraordinarily generous ­people. Their brains show distinctive patterns of activity. The difference is found specifically in the amygdala, a region of the brain that pro­cesses emotions. A study of kidney donors—an act of extraordinary altruism—­found that their amygdala was activated more than that of ordinary p­ eople when shown pictures of frightened p­ eople, and much more so than that of psychopaths, who actually show impairment in this region of the brain.24 In other words, altruists have a heightened empathic sensitivity.25 As for its origins, an older theory of altruism argued that helping ­others was beneficial to oneself ­because ­those one was likely to help ­were also likely to share one’s DNA, but this theory is no longer accepted by most scholars. The cooperative be­hav­iors described above cannot be explained by direct ge­ ne­tic relations since they occur among ­those who are not kin. Other forms of cooperation have a similar pattern. A detailed study of chimpanzee males in the wild, for example, found that they cooperate—by sharing meat, grooming, forming co­ali­tions, guarding territory—­mostly with unrelated individuals, rather than with b­ rothers or close relatives.26 Samuel Bowles and Herbert Gintis, who have done extensive research in the field of evolutionary social science, recently reviewed the latest theoretical issues. Based on extensive evidence and compelling logic, they conclude that h ­ umans evolved to care about ­others and cooperate with them b­ ecause groups that did so ­were more successful at protecting fellow group members from neighboring tribes during violent intergroup conflict.27 Other theorists

58 • Chapter 2

have proposed that preferences for fairness may have evolved among apes and then h ­ umans as a means to sustain cooperation among group members.28 And yet, it would be easy to exaggerate the extent to which ­humans are altruistic. The religious codes that evolved in­de­pen­dently across so many ­human socie­ties would not have been needed if the dominant and default ­human instinct was to submit oneself to the desires and benefits of the group. Despite the frequency of altruism, selfish be­hav­ior—­sometimes extremely selfish be­hav­ior—is also found across experimental settings in evolutionary social science. In a competitive environment with ­limited resources—an empire of scarcity—­individuals, f­ amily units, and groups are more likely to exhibit selfish be­hav­ior. The biologist E. O. Wilson has written about the evolutionary tension between selfishness and altruism and believes it has profound consequences for ­human morality and our sense of justice: Se­lection at the individual level tends to create competitiveness and selfish be­hav­ior. . . . ​In opposition, se­lection between groups tends to create selfless be­hav­ior, expressed in greater generosity and altruism, which in turn promote stronger cohesion and strength of the group as a ­whole.29 In fact, some degree of selfishness at the group level—­“parochialism,” in the scientific lit­er­a­ture—­may increase the advantage of within-­group altruism by increasing the fertility and survival rates of groups with members characterized by strong loyalty to one another and a willingness to fight outsiders, according to one prominent theory.30 Still, two crucial issues related to the viability of groups are: How are outside groups regarded, and how easily can the bound­aries between groups be bridged? Early ­human tribes that created unbridgeable divides between groups would have perished. Unwavering hostility t­ oward outsiders or a refusal to see them as equals would have made tribes vulnerable to an alliance of competing tribes. It would also have prevented trade between groups, without which many groups would have perished in the wake of disease, drought, or natu­ral disaster. Almost ­every hunter-­gatherer society ever observed was recorded to engage in trade across groups.31 Indeed, if h ­ umans w ­ ere incapable of including diverse ­people in their communities, then ancient and modern cities, and their intellectual, cultural, and technological achievements, would

Natu­ral Foundations of a Just Society  •  59

never have been pos­si­ble. Th ­ ere is clear evidence from wars that h ­ umans form alliances across ethnic and religious lines, even in spite of former or existing antagonisms.32 At a more mundane level, hundreds of studies have demonstrated that cooperative contact and geographic integration enhance trust and lower prejudice across racial and ethnic lines.33 Cicero dwelled on t­ hese considerations explic­itly in his treatise “On Duties.” Starting from his view of a common humanity, he laid out princi­ples for when to go to war—to avoid injury and maintain peace, never to profit at the expense of ­others—­and how to treat the vanquished ­a fter war: “­Those whom you conquer are to be treated kindly, t­ hose who, laying down their arms in good faith,” should receive food and shelter. He praised Romans from centuries past for admitting p­ eople from conquered foreign tribes as citizens in the aftermath of war, provided that they w ­ ere not “cruel and in34 human in war.” The task of just institutions is to set up rules to balance the complex, competing interests within and between individuals, families, and groups.

FROM ALTRUISTIC GROUP MEMBERS TO JUST SOCIE­T IES

With this background in mind, let’s explore what the fundamental conditions of a just society should be. Writing in the thriving Republic of Athens around 380 BCE, Plato set out to describe a just society. His starting point was to imagine what h ­ umans need at the most basic level, such as food, shelter, and clothing, and what system best meets t­ hese needs. Plato argues that h ­ uman needs can be met only through cooperation, which benefits every­one through mutually beneficial exchange and occupational specialization. ­People who have come together for ­these exchanges set up rules among themselves, forming a state. Anticipating modern economic theory, Plato observes that p­ eople are more productive—­can create higher quality goods and ser­vices per hour of effort—­when they specialize in one occupation rather than trying to supply all of their needs by themselves. The efficiency gains from cooperation result from learning and practice and the natu­ral diversity of h ­ uman talents. “­There are diversities of natures among us which are adapted to dif­fer­ent occupations,” Plato writes.35 Beyond meeting basic needs by facilitating exchange, the state should also provide public

60 • Chapter 2

goods such as wise leadership, security, protection of property, and education. For Plato, it is clear that the best way to provide this diverse array of goods and ser­vices is that “each individual should be put to the use for which nature intended him.” Drawing on the justice-­as-­proportionality princi­ple, Plato has Socrates suggest that “justice is the giving to each man what is proper to him.”36 What is due corresponds to what is needed, which is highly varied—­a fair interpretation of the law, food, shelter, shoes, medical care, and so on—­ and providing the precise t­ hing that is due is best accomplished by expert specialists. For Plato, it follows that justice is ­doing one’s own business, a job that suits one, and a just state is one in which ­people do the occupation that best suits them. This is how p­ eople in a community get what is due to them. In order to bring every­one up to his or her optimal level of productivity, Plato emphasizes the importance of education from an early age. While The Republic discusses the strengths and weaknesses of a hy­po­thet­i­cal curriculum and a few other details, Laws, a ­later and less famous work, elaborates on much more ambitious educational ideas. Th ­ ere, Plato endorses compulsory public education “open to all”—­that is, including both boys and girls, both of whom would receive military and academic training at the public’s expense. One passage suggests that he would start public education at around age three and continue ­until adulthood. “The most impor­tant part of education is training in the nursery,” he wrote.37 This policy makes him more progressive than many U.S. politicians who erroneously oppose public “preK,” as it is called, as an extravagant waste of money. Moreover, in The Republic and Laws, Plato explic­itly argues that w ­ omen’s “original nature is the same” as men’s. W ­ omen should be warriors, he claims, when they prove capable ­a fter preliminary training. “Many ­women are in many t­ hings superior to many men,” Plato posits, through the character of Socrates, whom he uses to convey his vision.38 To forbid naturally talented ­women from pursuing the occupations of men is a “violation of nature” and unjust. Consistent with his views on education, Plato considers that merit is not determined by social class and that pathways to leadership positions must be available to ­people of e­ very social rank. In ­these re­spects, Plato advocates for a society in which ability and merit, forged out of common access to public goods such as education and security, determine occupational status and positions of leadership.

Natu­ral Foundations of a Just Society  •  61

­ ese are compelling princi­ples worth preserving that follow logically Th from his unstated goal of maximizing the happiness of his ideal city-­state. Yet, Plato proves too willing to subordinate the individual to the state. In The Republic, he places a ­great deal of faith in the wisdom of rulers, even if chosen meritoriously, and gives them too much power to dictate the lives of ­people, including what they learn in school, whom they are allowed to reproduce with and marry, and even their access to property of any kind. In this sense, the citizens of Plato’s Republic, and even more so the slaves, are not respected as individuals who are capable of making even basic decisions about their own lives. ­These flaws are less extreme in Laws than in The Republic. In Laws, he is more respectful of republican institutions and more concerned about the concentration of power and wealth.39 For example, he praises three constitutional policies in Sparta that curbed the executive power of rulers.40 In the ideal city presented in Laws, Plato proposes a complex system of elections to ensure that all classes are po­liti­cally represented. He also proposes rather dramatic reforms to property in order to make wealth more equally distributed and curb what he regards as unfair advantages that flow to the rich.41 In the end, Plato’s vision of a just society should not be accepted in its entirety by any means. He argues that we should not look to enhance individual happiness but rather to aim for happiness that “resides in the state as a ­whole.” 42 Perhaps, this view explains why The Republic has many bad ideas about suppressing f­ree speech and artistic activity, eliminating property owner­ship among soldiers, using false propaganda to motivate public ser­vice and acquiescence, and forcing soldiers to share common wives and not know who their ­children are. Still, Plato’s ideas offer impor­tant princi­ples worth preserving: checks on po­liti­cal power, gender equality, the need for representative government, equality of access to occupations, and public universal education. ­These ideas support his fundamental vision, which is worth defending: A just society is one in which p­ eople do what they do best. John Rawls is arguably the Plato of modern philosophy, in that no single scholar has influenced academic views of justice more than he, at least in wealthy Western countries. His arguments are worth studying in detail. Unlike Plato, who begins by asking how society’s needs should be met, Rawls begins with an individual’s concerns and asks readers to engage in an imaginary exercise. It’s a bit like the following scenario: Pretend you are in a room with ten p­ eople you regard as your equals. All of you know that you

62 • Chapter 2

are g­ oing to compete and potentially cooperate with one another in a game, whose basic rules you have to work out together. The t­ hings you care most about—­life, health, dignity, income, happiness—­all depend on your relative per­for­mance in this game. None of you can know before you make the rules how your individual strengths and weaknesses ­will ­matter in this game, and so t­ here is no way to design the rules to take advantage of your specific abilities. What princi­ples would you work out to govern this game ­after rational deliberation? ­These background conditions are fair in the senses described above, as proportional, equal, and unbiased. Hence, the princi­ples that follow from this deliberative exercise are just. Of course, Rawls’s ­actual argument is not about a game. Rather, it concerns the “fundamental terms of association” among p­ eople and the institutions that govern be­hav­ior in a group of p­ eople. But it is similar to the exercise above in that he asks readers to set aside any knowledge about their own social status at birth or their natu­ral abilities. From this “original position,” Rawls makes a strong analytic case that a logical person would want a society characterized by two basic princi­ples (the second of which has two distinct parts): First Princi­ple: Each person has the same indefeasible claim to a fully adequate scheme of equal basic liberties, which scheme is compatible with the same scheme of liberties for all; Second Princi­ple: Social and economic inequalities are to satisfy two conditions: a. They are to be attached to offices and positions open to all ­under conditions of fair equality of opportunity; b. They are to be to the greatest benefit of the least advantaged members of society (the difference princi­ple).43 The radical equality of the original position—­where advantages of social status are wiped out—­encourages re­spect of o­ thers as oneself. That means giving them the freedom to learn, develop, and choose the occupation that suits their interests, ambitions, and natu­ral and acquired talents. It means allowing them to foster the familial and social relationships that enrich their lives and to en-

Natu­ral Foundations of a Just Society  •  63

gage in work that is meaningful to them. It means letting them keep much of the value of their work and use it as they see fit and cooperate and exchange with ­others when they believe it is in their benefit to do so. It also means empowering them to choose their po­liti­cal representatives, evaluate them, and hold them accountable not only to their own distinct priorities—­informed through experience, education, and citizenship—­but also to the public good. For Rawls, the first princi­ple includes ­these ­things. He explic­itly refers to the U.S. Constitution’s Bill of Rights and subsequent amendments as examples of basic liberties: freedom of expression, assembly, and religion; freedom from unreasonable government claims on property; freedom from slavery; and equality and due pro­cess before the law. Part “a” of the second princi­ple evokes equality of opportunity in a way that is consistent with Plato’s Republic and Laws. Both Rawls and Plato ­imagined a society in which ­people pursue the occupations that engage their talents, without restrictions placed upon them based on their gender or parents’ social status. The most controversial ele­ment to ­these princi­ples is the difference princi­ ple, but I think the controversy is rooted in a misinterpretation. Some have taken it to mean that the rich should be forced to redistribute their income ­until the income of the poor has reached its upper limit, given total national resources. That is profoundly mistaken. Rawls did not take a position as to what policies should be implemented to raise the welfare of the least off. Instead, he suggested that the well-­being of the poor might be maximized ­under a f­ ree market economy. A ­ fter all, in a poor and unproductive society, the least advantaged w ­ ill live in truly deplorable conditions. The optimal level of taxation, public goods spending, and re­distribution that would maximize the well-­being of the poor is an open question. Historically, the government owner­ship of the means of production (communism) has not served the least advantaged members of a society well, and yet, the legitimacy of some positive economic role for the state is indisputable. To take just one sphere of economic activity as an example, evidence from the United States suggests that access to public goods, income transfers, and related programs do effectively raise the welfare of the poor and the ­children born into disadvantageous conditions.44 This does not, however, imply that increasing the value and use rate of income transfers is always the most just policy. Depending on the specific circumstances of the individuals involved, ­there may be more effective ways to raise the welfare of the least off, such as training programs or ­mental health ser­vices.

64 • Chapter 2

Without minimizing the richness and sophistication of his analytic argument, I would suggest that Rawls’s difference princi­ple is r­ eally no dif­fer­ent at its core than the Catholic preferential option for the poor or even Cicero’s secular republican maxim: “Do the most for ­those who need the most.” 45 Rawls himself had once thought he would become a priest, so it is hardly surprising that his views would be consistent with Catholic social teaching.46 The late French phi­los­o­pher Paul Ricoeur has argued that Rawls’s difference princi­ple and the background conditions of equality in his original position are “equivalent to the Golden Rule applied to institutions.” 47 Rawls explic­itly rejects the common competing utilitarian ethical idea of “the greatest good for the greatest number” on grounds that “each person possesses an inviolability founded on justice that even the welfare of society as a ­whole cannot override. For this reason, justice denies that the loss of freedom for some is made right by a greater good shared by ­others.” 48 Rawls is debating modern scholars, but he might as well have been debating Plato. ­Because Rawls is not willing to sacrifice individuals to a collective, his difference princi­ple “2a” is the only one that is perfectly consistent with the princi­ples of Plato’s Republic. It is instructive to compare t­ hese views of justice with other widely held doctrines. Both Rawls’s and Plato’s theories of justice are at odds with essential communist princi­ples of justice. One, famously articulated by Karl Marx, is “From each according to his ability, to each according to his need.” 49 For many reasons, it is unrealistic to expect that the value of every­one’s work w ­ ill be identical. Not only does value differ by job, but even p­ eople with the same job vary in their per­for­mance. Hence, this princi­ple requires equal compensation for unequal work, which is fundamentally unfair from a reciprocal perspective. Experimental evidence from economists at Brigham Young University suggests that ­people accept some levels of in­equality when the distribution is based on merit but reject it other­wise.50 Another Marxist idea that is problematic for justice is the public owner­ ship of the means of production. This is difficult to reconcile with occupational choice, insofar as entrepreneurship, small-­business owner­ship, and even self-­proprietorship are explic­itly foreclosed. If the means of production ­were truly nationalized in e­ very way, which did not happen even in Soviet Rus­sia, website designers, con­sul­tants, plumbers, electricians, hair­dressers, taxi ­drivers, and other in­de­pen­dent workers would be forced to apply for employ-

Natu­ral Foundations of a Just Society  •  65

ment with the state. ­These in­de­pen­dent work arrangements ­were generally against the law in Soviet Rus­sia, as was the hiring of private employees, though ­there ­were many informal and formal exceptions.51 In general, most p­ eople ­were forced to be employees of the state, regardless of what­ever talent they might have for entrepreneurship or innovation or what­ever desire they might have to work in­de­pen­dently and set their own work schedule and fees. The communist vision of society also fails to maximize the welfare of the least off in at least one vital way: Through the elimination of merit-­based compensation, it fails to provide the means and incentives to discover and implement innovative ideas through entrepreneurship or other means. The worst off have benefited greatly from advances in science, technology, and business that have raised the quality of goods and services—­including healthcare—­ relative to their costs.52 Recognizing this, the Soviet Union did include merit-­ based compensation schemes, but given the prohibitions on private business owner­ship, ­there ­were nonetheless very strong incentives against risking one’s savings (or capital) through research and development and business development. In noncommunist socie­ties, this risk-­taking has led to many new technologies and business ventures that have benefited the least off. At the same time, a strength of the communist vision is its commitment to comprehensive public support for the least off, who have often suffered unnecessarily in affluent noncommunist socie­ties, including the United States, where low or even moderate income severely restricts access to quality housing and impor­tant ser­vices such as healthcare, transportation, and postsecondary training, despite a patchwork of both public and private sources of charity. Rawls did endorse both liberal socialism and property-­owning democracy as potentially consistent with his princi­ples, so long as occupations are freely chosen, po­liti­cal power is distributed equally among citizens, economic power is not concentrated to the point of blocking competition, and public goods are provided in such a way as to equalize opportunity at the start of life.53 Social democracies in Northern Eu­rope have managed to reject public owner­ship of the means of production, preserve entrepreneurship and innovation, and provide comprehensive support to ­those with the least resources, creating socie­ties that realize greater justice than ­either communist socie­ ties or ­those, like the United States, with relatively weak safety nets. Basic libertarian conceptions of justice are also problematic in that they ignore w ­ hether or not p­ eople start their lives on an equal basis or have equal

66 • Chapter 2

access to public goods. This indifference violates Plato’s goal of maximizing social welfare by allowing ­people to do what they do best, and it violates Rawls’s sense of fairness by allowing in­equality to be strongly founded on unfair advantages, such as parental social status and economic power. Fi­nally, libertarianism is also inconsistent with the difference princi­ple insofar as leading libertarian thinkers such as Robert Nozick and Frederick Hayek oppose publicly funded endeavors that give priority to the well-­being of the poor and thereby support unnecessarily depriving the least off of welfare.54 Libertarian society would leave any efforts to provide for education, poverty relief, and upward mobility to private collective action, where the incentives to do so are weak, volatile, and subject to arbitrary decisions. Fundamentally, it would put the least off in a position of subservience to the rich. One can imagine what this would look like by studying conditions in ancient Rome’s patronage networks, in which the poor and ­middle classes provided vari­ous ser­vices to the super-rich in exchange for regular payments. ­These ser­vices could as readily be criminal as military in nature. They also included educational ser­vices, which meant that teachers ­were viewed as something like a private charity; thus, they could be easily exploited, as well as being subject to any losses suffered by their patrons.55 It’s worth mentioning in this context that no society has ever achieved mass literacy without publicly funded education.56 The phi­los­o­phers Amartya Sen and Martha Nussbaum have extended Rawls’s vision by considering what conditions true equality of opportunity require.57 The core idea developed by Sen is that individuals should be given the capacity to create the conditions of a flourishing life. The practical implication of this is that equality of opportunity means more than just equal eligibility for occupations and income generation. As Nussbaum argues, for a society to be just, it must provide its ­people the capabilities to pursue a life of dignity, including health, education, and some degree of control over their po­liti­cal and material conditions.58 ­These strike me as logical extensions of the core ideas developed by Rawls and Plato. Another test of theories of justice is to see how they are regarded by extreme po­liti­cal groups. Other than committed libertarians, perhaps, it would be hard to identify a po­liti­cal group in the United States more likely to be hostile t­ oward Rawlsian princi­ples than the Alt-­R ight, and yet, when surveyed, a majority of p­ eople identifying as “alt-­right” express support for the basic princi­ple of equal opportunity, and just as many agree with the princi­

Natu­ral Foundations of a Just Society  •  67 0.6 0.5 0.4

50.5% 38.8%

39.4% 29.8%

0.3 19.5% 21.7%

0.2 0.1 0

Agree or strongly agree (5–7)

Neither agree nor disagree (4)

Disagree or strongly disagree (1–3)

We should work to give all groups an equal chance to succeed. We should do what we can to equalize conditions for different groups.

Figure 2.1. ​Agreement with basic principles of justice on a 1–7 scale among those who identify with the alt-right movement Source: Forscher, P. S., and N. Kteily. “A Psychological Profile of the Alt-­R ight.” August 10, 2017. Retrieved from osf.io/preprints.

ple of re­distribution in f­ avor of greater equality as oppose it (figure 2.1). This makes it hard to imagine how the libertarian rejection of equal opportunity could ever become a widely supported view. In summary, the Rawlsian vision of a just society is remarkably consistent with universally held cultural conceptions of reciprocity and justice and experimental evidence regarding altruistic be­hav­ior among primates and humans. It balances individual liberty with egalitarian concerns about fairness in opportunity, and it provides opportunities for merit to influence welfare, while stripping away arbitrary advantages and alleviating disadvantages. ­These attributes stand in contrast to competing visions of justice from Platonic utilitarianism, communism, and libertarianism. Having established what I believe the most basic rules of a just society consist of, I offer a glimpse in the next chapter of what would happen if an equal-­opportunity society ­were allowed to unfold.

3 Merit-­Based Egalitarianism

A central claim of this book is that well-­functioning markets—­ characterized by mutually beneficial exchange among po­liti­cal equals—­lead to egalitarian outcomes with re­spect to income and well-­being. This is pos­si­ ble ­because the skills and traits that lead to productive work are broadly distributed. The extreme in­equality that exists in the con­temporary United States and other countries is not the result of well-­functioning markets or vast differences in natu­ral ability. Rather, it results from po­liti­cal in­equality and corrupted markets. In an egalitarian market society, differences in market-­based compensation would largely be the result of differences in productivity among individuals. Productivity would vary less than the current distribution of income in many countries b­ ecause the rich are presently compensated well above their productive contributions, and in an egalitarian market society opportunities would be equal in basic ways: Poor ­children would be given adequate shelter and nutrition, educated from a young age in schools of similar quality to ­those attended by rich ­children, and geo­graph­i­cally and culturally integrated with their more affluent peers, so that neither rich nor poor ­children would view the other as alien or develop idiomatic values and beliefs based on ignorance of the other. Adults, moreover, would have access to education and training opportunities so that they could foster the dispositions and skills that arise from their specific ge­ne­tic and environmental makeup. This background would give nearly all p­ eople the ability to be productive at something that is valued in society. I want to emphasize that government institutions, combined with private charity, would complement and partially rectify market outcomes so as to further promote egalitarian conditions. Such redistributive arrangements are

Merit-­ Based Egalitarianism • 69

consistent with our natu­ral inclinations to show altruism t­ oward ­t hose in need and treat ­others with reciprocal concern. Still, the claim ­here is that nonmarket re­distribution w ­ ill be less necessary in a po­liti­cally egalitarian market-­based society than in one characterized by ubiquitous restrictions on mutually beneficial exchange. This is the opposite claim of most p­ eople on the po­liti­cal Left in the twenty-­ first c­ entury, who assume that f­ ree markets are the source of in­equality and that government institutions—­including state-­supported worker institutions such as ­unions and the collective owner­ship of companies—­are the only means available to counteract market injustice. My claim is also contrary to arguments on the po­liti­cal Right that income in­equality is the result of a competitive but fair market pro­cess. This chapter sketches some of the conceptual issues that arise when thinking about how ­people could distinguish themselves in a just society and provides evidence that the traits ­people need for success are broadly available, even ­under current conditions in the United States, where social advantages in upbringing differ widely across groups.

COMPENSATION AND PRODUCTIVITY

In this context, productivity means the value created by someone’s work. Value, in this sense, is complex and highly dependent on the work being done, but ­humans have learned to define, recognize, and appreciate value when they see it in the course of their interactions with ­others. Consumers have proven to be very good at identifying the valuable traits of highly complex goods and ser­vices, ranging from artistic works, repair ser­ vices, meals at restaurants, higher education, vacations, and automobiles. Companies such as Yelp, Rotten Tomatoes, Amazon, and Consumer Reports systematically collect and report ratings and reviews of a wide range of products, and in many cases, scholars have found that when assessed against other standards, t­ hese mea­sures are valid predictors of product quality.1 Even for a ser­vice as intangible as higher education, recent evidence shows that consumers of higher education provide subjective evaluations of their experience that strongly predict better outcomes for other students, such as their probability of graduation, obtaining an advanced degree, or earning higher incomes ­after graduation.2

70 • Chapter 3

Subjective evaluations of worker per­for­mance are also extremely common, particularly in large companies.3 But ­there is strong evidence to suggest that in this context ­there are often serious issues associated with reviewer bias. Employee evaluations that rely on objective per­for­mance metrics are associated with better management practices and firm per­for­mance. Firms are more likely to adopt performance-­based pay where ­labor markets are relatively un­regu­la­ted—­and payment is based on market mechanisms.4 Where employee pay and pay growth are tied to per­for­mance within firms,5 man­ag­ers are not simply trading on po­liti­cal power, except insofar as their po­liti­cal advantages affect their productivity. Th ­ ese advantages might include greater learning opportunities, or, for many men, not facing the cultural expectations and burden of childcare and other domestic responsibilities, which so often derail the ­careers of many w ­ omen. Compensation based on per­for­mance w ­ ill not lead to perfect equality. Rather, it w ­ ill lead to greater fairness in compensation than currently exists ­because it diminishes the impact of po­liti­cal power, and links pay closer to productivity. In a po­liti­cally egalitarian society, ­people ­will be given nearly equal opportunities to become productive. Of course, t­ here are limits to the extent to which compensation can be linked to productivity, especially when it comes to ser­vices that are partially or fully paid for by governments. Within a given public organ­ization, ­whether a school district, hospital system, or military branch, it is straightforward to design a system linking relative pay to productivity, but the amount of revenue available to distribute to workers w ­ ill depend on po­liti­cal ­factors, such as how the voters in a society and their po­liti­cal representatives value the ser­ vices being performed and how much they can afford to allocate to ­those ser­vices. Nonetheless, even in socially demo­cratic countries, most goods and ser­ vices are privately produced, and t­ here is a g­ reat deal of room for l­abor markets to function well, if they are allowed to. Moreover, even when goods and ser­vices are publicly produced, po­liti­cal egalitarianism demands that compensation be fair, which necessitates that per­for­mance be considered. The current practice in the United States of paying teachers based on their education credentials and experience may seem like an egalitarian policy, but it is unfair in a fundamental sense: It grants power to older and more experienced teachers regardless of ­whether or not they are actually deserving. ­There is very strong evidence that subjective teacher ratings—­based on research-­based

Merit-­ Based Egalitarianism • 71

rubrics—­predict a teacher’s ability to promote learning, and yet ratings of any kind are rarely used to influence compensation b­ ecause of the bizarre politics of education.6

WHEN MARKETS GENERATE LOW WAGES

One impor­tant concern that may be raised against linking pay to markets relates to workers who provide what are typically poorly compensated but socially valuable ser­vices, such as early childcare, elementary education, elder care, home healthcare, and social ser­vices. Th ­ ese ser­vices are recognized as socially valuable b­ ecause they embody widely shared social and ethical obligations to c­ hildren, the poor, the weak, and other groups who need the support of their fellow citizens. Other ser­vices that are essential to the smooth ­running of society but are not deemed socially valuable in the same sense, such as cleaning ser­vices, food preparation, delivery ser­vices, and retail sales jobs, also pay consistently low wages. In addition to their social value, the provision of t­ hese ser­vices f­ rees ­others to do more specialized work of ­greater market value. Yet, consistently low pay suggests that markets bestow ­little value on ­these basic ser­vices. The providers face extreme competition, which surely holds compensation down. Most of the population could perform ­these tasks if they had to, and the providers of the ser­vices are competing against the buyers of the ser­vices themselves. For instance, the parents of young c­ hildren and the adult c­ hildren of el­ derly parents can provide educational or caretaking ser­vices to their ­family members, but may decide that it is more efficient or satisfying for them to do something ­else with at least a large portion of their time—­like hold a full-­ time job—­than devote most of their waking life to taking care of ­others for no pay. As a result, early childhood caretakers and educators have to compete against parents. If the price becomes too high, parents w ­ ill find it impractical or unaffordable and elect to quit their job and provide their own childcare. Similar competition holds down prices for cleaning ser­vices and restaurants. Most p­ eople are capable of preparing their own meals or cleaning their own living space, but if they are affluent enough, they may decide to hire other ­people to provide ­these ser­vices. As a result, ­there ­will always be a subset of occupations that pay a small percentage of the average worker’s wage. The economist Alan Manning has developed a formal analy­sis of

72 • Chapter 3

t­ hese dynamics showing that demand for low-­wage work ­will grow alongside the local income growth of the affluent.7 Other ­labor markets do not face this kind of pressure. Most ­people would find it impractical if not impossible to manufacture t­ hings they buy—­like a pencil, clothing, home, or new technology. ­Unless they live on a farm, it is highly unlikely that they could produce even a modicum of the food they need to consume. Likewise, for professional ser­vices: In law, finance, medicine, engineering, architecture, computers, and higher education the product is complex and specialized enough such that competition between providers sets prices, rather than competition between buyers and sellers. The same would be true for entertainment ser­vices. ­People ­don’t pay to watch amateurs play backyard football games or sing karaoke at bars, but they do pay to see elite athletes and musicians perform. Within low-­paying occupations, ­there is considerable variation in pay, and more productive workers may be rewarded, but even highly effective early childhood educators, nannies, and home healthcare workers often get saddled with pay rates that would fall below the levels of the worstperforming professional workers. They are also underpaid relative to their cognitive and noncognitive skill set, as I discuss l­ ater. Redistributive policies—­ such as income supplementation and the provision of benefits from the public—­are an appropriate response to this prob­lem for several reasons. Such forms of re­distribution are inherently fair relative to market distribution, ­because they come closer to accurately compensating ­people for the application of their skills and abilities as well as their real contribution to society.

PRODUCTIVE TRAITS, INCOME, AND HEALTH

In practice, per­for­mance is based on a set of skills and traits that are widely distributed now and would be even more broadly available u­ nder egalitarian po­liti­cal regimes. For example, in 1979, the U.S. Bureau of ­Labor Statistics (BLS) started collecting data from a random sample of 12,682 Americans born between 1957 and 1964 and continued to do so u­ ntil 2012. In 1981, it mea­sured cognitive ability with the Armed Forces Qualification Test (AFQT), an achievement test similar to college admission exams. Researchers at the BLS also

Merit-­ Based Egalitarianism • 73

collected results of traditional IQ mea­sures, such as the Stanford-­Binet, as well as a ­simple battery of five questions meant to capture the “Big 5” personality traits: extroversion, openness to new experiences, anxiety, self-­discipline, and agreeableness. Additionally, the survey mea­sured life outcomes such as ­family income and health status. An impor­tant research question is: To what extent do cognitive ability and personality traits predict income and health? The answer is quite a lot. But the answers would surprise ­those who believe that IQ is the only t­ hing that ­matters in earnings or life success. To study this I calculated average income in ­middle adulthood and compared it to cognitive and personality traits mea­sures in late adolescence. First of all, in terms of earnings, the achievement test of the AFQT is a notably better predictor of income than classic mea­sures of IQ.8 If both are in the same model predicting income or health using the 1979 National Longitudinal Survey of Youth (NLSY), only the AFQT is significant. This is not surprising. IQ tests are more abstract and less relevant to real-­world prob­lems than the AFQT and other mea­sures of cognitive achievement. A one standard deviation in cognitive ability, as mea­sured by the AFQT, predicts 22 ­percent higher f­ amily income, a­ fter controlling for other demographic f­ actors, education, and personality, whereas more traditional IQ mea­sures predict only a 9 ­percent increase. The AFQT is also more predictive of educational attainment than traditional IQ mea­sures. To put it bluntly, achievement tests are a better mea­sure of cognitive ability than conventional IQ tests, even though they are highly correlated with one another.9 Second, educational attainment is a stronger predictor of ­family income and health status than cognitive ability. In other words, when both ­factors are used as in­de­pen­dent variables, educational attainment has stronger statistical power (a higher t-­statistic). If you take two p­ eople with the same cognitive ability, the first person earns 10 ­percent higher ­family income for ­every additional year of study and reports significantly higher levels of general health. Third, personality traits—as mea­ sured by responses to just three questions—­are as impor­tant or almost as impor­tant as cognitive ability in explaining income and more impor­tant than cognitive ability in explaining health status. The questions that have the most predictive power are ­those that ask how p­ eople would rate themselves on three characteristic traits: extroverted/enthusiastic, dependable/self-­disciplined, and anxious/easily upset,

74 • Chapter 3

­ able 3.1.  Statistical relationship between cognitive ability, T personality, and other ­factors in predicting income and health in the United States

Educational attainment in years Cognitive ability Personality (3 out of “Big 5” traits) Parental education (mean of both parents) Perception of self-­control

­People born 1957–1964

­People born 1980–1984

­ amily F income, 2000–2012

Health

Personal income, 2013

Health

20.6 15.4 10.9 5

9.2 5.4 10.4 1.4

11.2 6.6 6.7 1.2

6.3 2.5 12.9 3.6

−3.3

−4.2

Notes: The first two columns are from the author’s analy­sis of NLSY 1979. The third and fourth columns analyze the NLSY 1997. All values are t-­statistics from regression of age and w ­ hether the respondent is Jewish, Catholic, another non-­Christian religion, black, or Hispanic on ­either the log of f­ amily income, personal income, or health status, on a 1–5 scale, with 5 being excellent and 1 being poor. Cognitive ability is mea­sured by the Armed Forces Qualification Test for the 1979 cohort and the Computer-­Adaptive Armed Ser­vices Vocational Aptitude Battery for the 1997 cohort, using mathematical knowledge (MK), arithmetic reasoning (AR), word knowledge (WK), and Paragraph Comprehension (PC) subtests. Personality is mea­sured by responses to three questions about the extent to which the individual agrees on a 1–7 scale with descriptions of him-­ or herself as extroverted/enthusiastic, dependable/self-­d isciplined, and anxious/easily upset. The personality score adds extroverted with dependable and subtracts anxious. Personality and cognitive ability are mea­sured in 1979 and 1997 for the respective surveys. Health status is asked around age 50 for the NLS 1979 cohort and age 30 for the NLS 1997 cohort.

with the last trait negatively predicting income and health and the other two being positively related. Likewise, when ­people report feeling that fate and luck control their lives, they tend to earn less income and be less healthy than ­those who feel a greater sense of agency. Personality is far more impor­ tant than even parental education. Fourth, the main analy­sis presented ­here assumes that income and cognitive ability move up in a linear relationship, such that a l­ittle bit higher IQ ­matters as much at the top as it does at the bottom. The more complex real­ ity is that changes in IQ m ­ atter less at the top than at the bottom. Cognitive per­for­mance maximizes income at about the 88th percentile. A ­ fter that, t­ here is no income benefit of having a higher IQ for the average person. The super-­ rich are no more intelligent than the merely affluent.10

Merit-­ Based Egalitarianism • 75

In sum, cognitive ability is not as impor­tant as educational attainment to income and health, and very rough personality mea­sures are equally impor­ tant for income and even more so for health. In other words, self-­discipline, enthusiasm, and the ability to avoid getting anxious and upset in the face of stress ­matter about as much as intelligence in terms of predicting success in life. Merit Pay

What are the implications of t­ hese findings in terms of access to the good ­things in life—­such as income and health? The argument ­here is that IQ, personality traits, and education are far more evenly distributed than the current distribution of income and health status would suggest. In a po­liti­cally egalitarian world, roughly equal access to a healthy ­family life and good schooling would even out IQ, personality differences, and educational attainment, but even in the unequal world we live in, paying p­ eople based on ­these characteristics—­instead of ­these characteristics plus po­l iti­cal power and market distortions—­would reduce income in­equality dramatically. To illustrate the point, one can compute existing in­equality in health status and income in­equality using the U.S. Bureau of ­Labor Statistics’ NLSY 1979 sample. What would income and health look like if they w ­ ere based only on IQ and personality traits? ­People on the po­liti­cal Right might assume that payment based on IQ and personality would generate greater in­equality, b­ ecause social and po­ liti­cal pressures—­such as regulation, ­unions, and affirmative action—­ overcompensate unproductive workers. Meanwhile, p­ eople on the po­liti­cal Left might assume that IQ and personality traits are like luxury goods that are purchased via a privileged upbringing and that paying ­people based on ­these traits would exacerbate inequalities that start in childhood. Both expectations would be wrong. In­equality in health status and income would be 34 ­percent and 56 ­percent, respectively, of the current distribution if based on “merit,” as mea­sured by cognitive ability and personality. In other words, in­equality in health would fall by two-­thirds and in­equality in income would fall by half. The United States would become one of the most egalitarian countries in the world if the only ­things that mattered to income ­were age, personality, and cognitive ability.

76 • Chapter 3 0.60 0.50

0.50 0.44

0.40

0.34

0.37

0.30

0.25 0.17

0.20 0.10 0.00

Actual inequality

Inequality if based on cognitive ability alone Health status

Inequality if based on cognitive ability and personality

Family income

Figure 3.1. ​Current U.S. inequality index by income and health and an alternative estimate if income and health were based on the distribution of cognitive ability and personality Notes: National Longitudinal Survey of Youth 1979 sample. Cognitive ability is mea­sured by the Armed Forces Qualification Test. Personality is mea­sured by responses to three questions about the extent to which the individual agrees on a 1–7 scale with descriptions of him-­or herself as extroverted/enthusiastic, dependable/self-­disciplined, and anxious/easily upset. The personality score adds extroverted to dependable and subtracts anxious. The analy­sis regresses AFQT score and then AFQT score and personality on health status at or near age 50 (mea­sured on a 1–5 scale) and ­family income (averaged from 2000–2012 and adjusted for inflation). The regression also controls for year of birth. The predicted values from this regression are used to calculate the Gini coefficient of in­equality, which is compared to the unadjusted Gini coefficient shown on the far-left bars. The sample size is approximately 6,000 to 8,000 adults born between 1957 and 1964.

Beyond the United States

The impor­tant but ­limited relationship between cognitive ability and earnings applies to developed countries beyond the United States. In 2011, the Organisation for Economic Co-­Operation and Development (OECD) implemented the Programme for the International Assessment of Adult Competencies (PIAAC) to mea­sure cognitive ability of adults aged 16 to 65 around the world and better learn how t­ hose skills relate to ­labor market and health outcomes. Detailed income data and cognitive test scores are available for 78,000 individuals in 23 countries, including the United States, in the main public

Merit-­ Based Egalitarianism • 77

data files.11 I mea­sure cognitive ability by combining standardized scores on math, literacy, and technology exams. A standard deviation in cognitive ability predicts a 13 ­percent increase in earnings. That is a substantial effect, and it is highly significant (meaning the margin of error is low), but just as with the U.S. data discussed above, it is comparable to the effect of educational attainment, work experience, and gender. In fact, work experience mea­sured by age has slightly more predictive power for earnings than cognitive ability.12 Education and gender are also slightly stronger predictors of income than cognitive ability, while health status is slightly weaker. The PIAAC does not ask about personality directly, but it does ask ­whether the respondent likes learning new ­things, is able to relate new ideas to real life, and ­whether he or she trusts other ­people. Positive answers to ­these questions are roughly as predictive of earnings as cognitive ability and remain highly predictive when controlling for cognitive ability. Likewise, higher cognitive ability is a strong predictor of health status across countries, but t­hese f­actors—­education, trust of o­ thers—­are just as good at predicting health. As in the U.S. case, if pay w ­ ere strictly tied to cognitive ability, as well as education, gender, and age, income distribution would become far more egalitarian. Health status would also be more egalitarian. In­equality in health status, as mea­sured by the Gini coefficient, would fall from 0.52 to 0.21. In­equality in income would fall from 0.47 to 0.17.13 The reason is that in nearly ­every rich country elite professionals and man­ ag­ers are paid the highest salaries, but ­those salaries exceed what would be predicted based on the cognitive ability and other mea­sured attributes of the workers—­such as age, education, gender, and the education of their parents. This earnings advantage of working in a managerial or professional occupation is 28 ­percent for the average developed country in the PIAAC database, but it is especially pronounced in highly unequal countries such as Chile, Israel, the United Kingdom, and the United States (exceeding 37 percent), while it is much lower in more egalitarian Scandinavian countries and the Netherlands. Paying ­people based on cognitive ability would be redistributive. To put the relatively egalitarian nature of education in perspective, consider that the richest 1 ­percent of Americans account for roughly 20 ­percent of income but just over 1 ­percent of years of education because the average rich person is only slightly more educated than the average nonrich person.14 A millionaire (in U.S. dollars) earns somewhere around 20 times the typical OECD worker’s income (roughly $50,000), but it is not ­because the

­ able 3.2.  The earnings advantage (%) of working in a professional T occupation above cognitive ability, education, and other characteristics, by country, 2011–2015

Chile Japan New Zealand United Kingdom United States Ireland France Israel Netherlands Poland Norway Czech Republic Slovenia Lithuania Spain ­Korea Italy Denmark Belgium Slovak Republic Greece Rus­sian Federation

All professionals

Man­ag­ers

Professionals excluding managers and technicians

0.59 0.43 0.43 0.39 0.37 0.36 0.34 0.32 0.32 0.30 0.24 0.22 0.21 0.19 0.19 0.18 0.18 0.18 0.17 0.14 0.08 −0.07

0.86 0.73 0.58 0.54 0.56 0.56 0.68 0.52 0.44 0.56 0.44 0.52 0.39 0.27 0.35 0.49 0.27 0.46 0.26 0.04 0.36 0.17

0.72 0.39 0.39 0.24 0.38 0.35 0.37 0.28 0.26 0.26 0.16 0.22 0.23 0.22 0.17 0.16 0.12 0.12 0.14 0.07 0.00 −0.22

Technicians and associate professionals 0.45 0.36 0.32 0.33 0.28 0.26 0.27 0.24 0.29 0.22 0.20 0.14 0.16 0.14 0.15 0.13 0.19 0.18 0.16 0.23 0.11 0.02

Note: Author anlaysis of data from OECD PIAAC. Professional premium is the average percentage difference in adjusted earnings for those working in a professional occupation compared to nonprofessional workers. The second column (“All professionals”) is the coefficient or slope of a linear regression of the log of earnings on cognitive ability—measured as the international standardized values of mean scores on math, literacy, and technology; indicators for education (including primary, associates, bachelor’s degree, and graduate degree); age measured in quartic values (age, age2 , age3, age4); a binary variable for male; whether respondent is born in the country; whether respondent is a native language speaker; country fixed effects; and a measure of social advantage. The social advantage indicator is the predicted value from a regression of mean cognitive ability on the number of books in home as child; parental education; gender; age; whether born in country or not; and whether a native language speaker. The results in column 2 are from one regression. Results in columns 3–5 are from a separate regression where the three professional groups are analyzed independently, with the omitted category being nonprofessionals. To interpret one line, in Denmark, professional workers earn just 18 percent more than nonprofessional workers—compared to 59 percent in Chile—after adjusting for cognitive ability, education, parental background, and other factors. Elias, P. (1997), “Occupational Classification (ISCO-88): Concepts, Methods, Reliability, Validity and Cross-­National Comparability,” OECD ­Labour Market and Social Policy Occasional Papers, No. 20, OECD Publishing, Paris. http://­d x​.­doi​.­org​/­10​.­1787​/­304441717388.

Merit-­ Based Egalitarianism • 79

millionaire’s IQ is 20 times higher. An IQ score at the top 99.9th percentile (149, or 3.3 standard deviations above the mean) would translate into predicted earnings of $76,000.15 Indeed, the very small number of ­people who took the PIAAC exam and scored two standard deviations above the mean (which would mean an IQ over 130) exhibited the same average earnings and health status as t­ hose who score between 115 and 130. The richest 1 ­percent of income earners globally in the PIAAC database have a mean IQ score of roughly 108.8, which is high but not extraordinary. It is below that of the average person with a postgraduate degree (109.5). The United States scores somewhat low compared to other countries in the PIAAC database, so it should not be surprising that the top 1  ­percent of income earners in the United States score below top earners globally, exhibiting a mean IQ of 103.6, despite being richer. The PIAAC database reports cognitive ability results for 540 managing directors and chief executives and 339 other occupations with at least 20 observations globally. CEOs are among the highest paid occupations, as you would guess. To account for country differences in compensation, I use the average decile rank for income within each country. The median CEO in the database is at the 9th decile (meaning that he or she is paid more than 80 ­percent of the population of that country). ­There are 195 occupational categories that score above managing directors and CEOs on cognitive per­for­mance. The average CEO scores about 0.3 standard deviations above the mean adult, which corresponds to an IQ of 104.7. By comparison, the average African American with a postgraduate degree in the PIAAC database has an IQ of 103. Stated other­wise, the average African American who has gone to gradu­ate school has roughly the same cognitive ability as the average CEO living in 29 developed countries. The average CEO is overpaid by roughly 23 ­percent relative to his or her cognitive ability, education, age, and other skills mea­sured in the database. It is likely that CEOs have personality traits, intangible skills, and personal connections that are valuable to companies, but their cognitive ability does not appear to be among their main assets. Likewise, ­lawyers and ­legal professionals are overpaid relative to their cognitive ability. In Norway and the UK, PIAAC classifies ­lawyers as “­Legal professionals.” Their mean cognitive ability score is extraordinarily high (119.4 or 1.3 standard deviations above the global mean for all 29 rich countries in the database), but they are still overpaid by 62 ­percent compared with p­ eople

80 • Chapter 3

with identical cognitive scores, levels of education, and other characteristics. In the United States, l­egal professionals have a cognitive score of 115 (on a global basis—so slightly lower than ­lawyers in the UK and Norway) and excess compensation of 23 ­percent. Elsewhere in rich countries, t­ hose classified as “­Lawyers” score much closer to the average worker (109.6), slightly below a draftsperson (someone who makes detailed technical drawings), and well below many other professional occupations. This group is overpaid by 17 ­percent on average, while the average draftsperson, who typically does not have a four-­year college degree, earns a premium of only 7 ­percent. Medical doctors and specialist medical prac­ti­tion­ers score at roughly the same level as l­awyers across countries (108.4 or 110.4 in the United States), but are not overpaid on average in the 20 countries outside of the United States (mean premium is −.02 ­percent), likely b­ ecause of the central role of governments in setting doctor pay in t­ hose countries. In the United States, medical students and residents in training may not be overpaid, but medical doctors over the age of 40 bring in a salary premium of 48 ­percent. Incidentally, dentists score slightly below doctors (107.2), but are paid a premium of 38 ­percent across all countries in the database. In general, the highest-­scoring professional groups on cognitive exams are overpaid relative to their objective scores and characteristics, while low-­scoring workers in jobs that require l­ ittle training are underpaid. Moreover, life science professionals, software application developers, university and higher-­ education teachers, and social and religious professionals all exhibit high cognitive scores (111 or higher across 29 countries), but they are not paid nearly as well as most man­ag­ers, financial analysts, or ­lawyers. Sociologists and anthropologists, for example, score the same on cognitive tests as financial analysts (112), but their salaries are 77 ­percent below the average worker who has similar skills, whereas financial analysts have salaries 46 ­percent above comparable workers. More broadly, workers in the lowest-­skilled jobs—­food preparation, cleaners, and helpers—­are underpaid rather substantially, taking into account cognitive ability and other traits, whereas professional man­ag­ers tend to be the most overpaid. Information and communications technology professionals are an exception in that they have high cognitive scores but are paid in line with ­those skills (table 3.3). Any way you slice the data, paying p­ eople based solely on cognitive ability would vastly reduce income in­equality compared to the current distribution.

Merit-­ Based Egalitarianism • 81

And cognitive ability is more unevenly distributed than the personality traits conducive to higher pay. Likewise, training and experience are likely to be more evenly distributed than cognitive ability, and all of t­ hese f­ actors could be spread more evenly with policy changes, as I w ­ ill discuss in l­ ater chapters.

RACIAL AND GENDER DIFFERENCES IN VALUABLE SKILLS ARE SMALL, AND MALLEABLE ENVIRONMENTAL MECHANISMS ARE READILY IDENTIFIED

As demonstrated above, IQ and valuable personality traits are distributed more equally than income and health status in the United States, as mea­sured by the Gini coefficient of in­equality.16 Moreover, the importance of IQ for income and health is not nearly as dominant as some scholars have implied. ­Factors such as age, participation in ­labor markets, and personality are more impor­tant when combined. Cognitive Skills

With re­spect to racial and ethnic differences on standard cognitive per­for­ mance mea­sures, con­temporary data show that the average scores for black ­people and Hispanics are lower than for non-Hispanic white ­people and Asians. This is the case for adolescents who w ­ ere born in the United States between 1980 and 1984 and tested with the Armed Ser­vices Vocational Aptitude Battery (ASVAB), according to my analy­sis of the 1997 NLSY. It also applies to data in the OECD’s PIAAC examination of adults.17 The black-­white gap in the NLSY 1997 cohort is roughly 0.86 standard deviations, which is slightly lower than the gap reported by Charles Murray and Richard Herrnstein in The Bell Curve, their influential and controversial book about intelligence.18 The gap between Hispanics and non-­Hispanic white p­ eople is 0.63 standard deviations in the NLSY 1997 and slightly more in the PIAAC. In the NLSY, Jews and Asians score higher than white Christians or white ­people on overall cognitive per­for­mance, but the sample sizes are small and the differences are not significant ­a fter adjusting for parental education. White Christians from highly educated families score just as highly on IQ tests as their Jewish and Asian peers from similar backgrounds in ­these data.

Mean cognitive score 114.5 111.3 110.8 110.6 109.5 109.4 108.8 108.0 107.4 107.4 107.3 105.7 105.6 105.4 104.3 104.2 103.1 102.9 102.4 102.3 102.3 102.2

Occupation group

Information and communications technology professionals Business and administration professionals Administrative and commercial man­ag­ers Science and engineering professionals Technicians and associate professionals Information and communications technicians ­Legal, social, and cultural professionals Production and specialized ser­vices man­ag­ers Teaching professionals Health professionals Chief executives, se­nior officials, and legislators Business and administration associate professionals Armed forces occupations General and keyboard clerks Electrical and electronic trades workers Science and engineering associate professionals Customer-­services clerks Numerical and material recording clerks Hospitality, retail, and other ser­vices man­ag­ers Health associate professionals ­Legal, social, cultural, and related associate professionals Other clerical-­support workers

9 9 9 8 6 7 6 9 7 7 9 7 6 5 6 8 4 6 7 5 5 5

Median income decile −3% 29% 49% 14% 17% 7% −4% 37% −10% 4% 25% 20% 14% −4% 11% 15% −6% 7% 32% 4% −11% 6%

(continued)

Mean income premium

­ able 3.3.  Cognitive per­for­mance, rank in income distribution, and income premium by occupation across T 29 developed countries, 2011–2015

5% −4%

6 4 4 3 2 2

93.1 90.4 90.2 87.5

−51% −19% −31% −31%

8% −27% −5%

5 3 6

95.8 95.6 94.9 94.9 94.7 93.9

−7% −14% 2% 12% 6% −13% −19% −6%

6 3 5 6 5 3 3 4

101.4 100.1 99.7 99.2 96.5 96.5 96.0 95.8

Source: PIAAC database for 29 countries and 188,168 adults aged 16–65. Median decile of income is shown, using within-­country income rank. Income premium is calculated by regressing the log of ­actual income (available for seven countries in main data files) on cognitive ability—­measured as the international standardized values of mean scores on math, literacy, and technology, indicators for education (including primary, associates degree, bachelor’s degree, and gradu­ate degree), age mea­sured in quartic values (age, age2 , age3, age4), a binary variable for male, ­whether respondent is born in the country, ­whether respondent is a native-­language speaker, country fixed effects, and a mea­sure of social advantage. The social advantage indicator is the predicted value from a regression of mean cognitive ability on the number of books in home as child, parental education, gender, age, ­whether born in country or not, and native-language speaker.

Handicraft and printing workers Sales workers Protective-­services workers Metal, machinery, and related trades workers Assemblers Personal-­service workers Personal-­care workers Food pro­cessing, woodworking, garment, and other craft and related trades workers Stationary plant and machine operators Refuse workers and other elementary workers ­Drivers and mobile plant operators No paid work for past five years Building and related trades workers, excluding electricians Laborers in mining, construction, manufacturing, and transport Market-­oriented skilled agricultural workers Agricultural, forestry, and fishery laborers Food-­preparation assistants Cleaners and helpers

84 • Chapter 3

The PIAAC sample is much larger but does not collect data on religion. It shows that U.S.-­based Asian and Pacific Islanders score lower than non-­ Hispanic white ­people by roughly one-­third of a standard deviation, despite higher levels of education. This is based on an overall score, but white ­people also score higher on each of the three test sections—­math, literacy, and technology. Upcoming chapters w ­ ill discuss the nonge­ne­tic origins for t­ hese gaps and how they have changed—or even reversed—­over short periods of time. Despite the gaps between racial and ethnic groups, ­there is enough overlap such that substantial fractions of the black, Hispanic, and working-­class white population already score high enough to enter and succeed in e­ very profession. For example, in the United States, 25 ­percent of information and communications technology professionals score below 105 on the PIAAC (using U.S.-­based means and standard deviations). That is the highest-­scoring broad occupational group in terms of educational attainments (83 ­percent have at least a bachelor’s degree). Yet, more than one in five African Americans with a bachelor’s degree score above that entry-­level threshold in cognitive ability. Likewise, a sizeable percentage of information and communications technology professionals in the United States are black (7  ­percent in the PIAAC database). For now, the impor­tant point to take away is that black ­people, w ­ omen, and ­people born to less educated ­mothers are all currently paid below what their cognitive scores alone would predict, as suggested by data from the NLSY 1997 and the PIAAC.19 In other words, pay would be more egalitarian by race, class, and gender, if ­people ­were compensated based purely on formal mea­sures of cognitive ability. A ­later chapter of this book ­will discuss one reason why this is the case: Elite professionals, who are often white males from educated backgrounds, create policies and cultural norms that prop up their incomes. ­There are other ways to think about cognitive ability that go beyond the abstract reasoning and literacy skills that IQ and achievement tests are designed to mea­sure. Th ­ ese include musical, artistic, or per­for­mance ability, though I know of no systematic data collection effort to mea­sure ­those skills and how they are distributed. ­There have been efforts to mea­sure cognitive skills related to skilled blue-­ collar work. The military vocational exam tests nonacademic technical knowledge about automotives, shop or craft work, electronics, and mechanics, and

Merit-­ Based Egalitarianism • 85

t­ hese data are also available in the NLSY 1997. Black ­people score no lower than white ­people on automotive or shop knowledge, and Asians, white Christians, and Jews all score the same on all four mea­sures. Mastering this body of technical knowledge can be helpful for a variety of skilled trades in construction, repair, and production. While the vocational test is not intended to mea­sure IQ, the fact that black p­ eople perform the same as Jews, Asians, and white ­people on at least two of the four subtests illustrates the power of environmental exposure and influence to affect test scores. So-­called hereditarians argue that IQ mea­sures genet­ically determine a general intelligence that is applicable to any intellectual prob­lem, but repairing machines is a major intellectual challenge, and ­these data suggest that ­people with somewhat low IQs—as typically measured—­may be just as good as p­ eople with higher IQs at solving this type of prob­lem. In fact, racial gaps in knowledge are much narrower when it comes to the technical and scientific knowledge that ­people commonly use in real-­world jobs. The U.S. Bureau of ­Labor Statistics surveys workers in ­every detailed occupational category (roughly 900 of them) about the level and use of science, technology, engineering, and math (STEM) knowledge in vari­ous domains through its O*NET program. ­These domains include science (biology, chemistry, physics), engineering, computers, math, design, mechanics, construction, production, economics, and telecommunications. Expanding an analy­sis that I conducted for the National Academies of Science using ­these data, I find that black p­ eople work in occupations that are just .24 standard deviations less demanding of STEM knowledge or technical knowledge than white workers. The average Hispanic worker is in a job that requires .20 standard deviations less STEM knowledge and a similar amount of more general technical knowledge.20 The broad occupational groups with the highest scores on STEM knowledge are engineers, healthcare prac­ti­tion­ers, scientists, computer workers, man­ag­ers, installation, maintenance and repair workers, and construction workers. The cognitive skills needed to perform t­ hese jobs are clearly valuable in that workers with the same level of education and experience earn higher incomes if they report higher knowledge in ­these domains. The magnitude is such that a one standard deviation in technical knowledge (or STEM knowledge) predicts 13 ­percent (or 15 ­percent) higher incomes. And yet, knowledge in ­these technical domains is commonly held across all groups of ­people even in a highly unequal country such as the United States.

86 • Chapter 3

The Unnatural Gaps in Scores by Gender

One might also be concerned about gender differences in cognitive skills. In 2005, the economist Larry Summers, who also served as the president of Harvard University and U.S. Trea­sury secretary, infamously claimed that the overrepre­sen­ta­tion of men among top scientists was at least partly the result of the “dif­fer­ent availability of aptitude at the high end,” by which he meant that t­ here are simply not enough ­women with very high IQs to be top scientists and that biological ­factors explain this. In response to a question ­after his speech, he clarified his remarks: My point was simply that the field of behavioral ge­ne­tics had a revolution in the last fifteen years, and the principal thrust of that revolution was the discovery that a large number of ­things that p­ eople thought ­were due to socialization w ­ eren’t, and w ­ ere in fact due to more intrin21 sic h ­ uman nature. The phenomenon that Summers was trying to explain is why only 19 ­percent of U.S. PhDs in physics are awarded to w ­ omen; shares for computer science (21 ­percent), engineering (23 ­percent), and mathe­matics (29 ­percent) are only slightly higher.22 It is not difficult to test and reject Summers’s hypothesis, which harkens back to the erroneous assumptions about w ­ omen made by male intellectuals 23 throughout recorded history. One of the largest-­ever studies of IQ scores among high school students in the United States—­called Proj­ect Talent—­was undertaken in the 1960s, when gender norms ­were particularly unfavorable to ­women. Young ­women born in the 1940s had the same mean IQ as young men but comprised only 35  ­percent of top-­scoring students, defined as t­hose scoring at least two standard deviations above the mean (IQ of 130 or more).24 Only 33 ­percent of top-­scoring students ­were female for math and only 13  ­percent for science. Yet, much has changed culturally since the mid-­twentieth ­century. International data on adolescents born at the start of the twenty-­first c­ entury show that young w ­ omen have made g­ reat pro­gress in catching up to the highest-­ scoring young men. On the OECD’s Programme for International Student Assessment (PISA) exam of 15-­year-olds, 44 ­percent of top-­scoring

Merit-­ Based Egalitarianism • 87

test-­takers—­those at least two standard deviations above the mean—­are currently ­women, for overall scores across all subjects.25 This is true for both the international average and the United States. By subject, ­women comprise 37 ­percent of top scorers for math, 39 ­percent for science, and 60 ­percent for reading. The breakdown is similar in the United States.26 Cultural issues clearly come into play. Girls, but not boys, score lower on science and math when their m ­ others have lower levels of schooling than their f­ athers.27 In 25 of the 73 geographic units (a mix of countries and regions within countries) with test score data, the average girl outscores the average boy on math. It is difficult to imagine a biological explanation for the variation observed across countries or the emergence of more females among top scorers over the span of a few de­cades within the United States, from 13 ­percent to 41 ­percent in science. Noncognitive Skills

While cognitive skill gaps between racial, gender, and ethnic groups are well documented, t­ here are no black-­white differences overall or Hispanic-­white differences in personality assessments; moreover, on the personality traits that predict health and income status described above on the NLSY 1997, Jewish-­ Americans have the same scores as Christian-­Americans, and Asian-­Americans score only slightly lower than white p­ eople. Black p­ eople are significantly more likely than white ­people to report being enthusiastic and self-­disciplined, both traits that predict higher income and health, but black ­people are more likely to report being anxious, which predicts lower income and health. Asians rate themselves the same as white ­people on self-­discipline and anxiety, but lower on enthusiasm. Environmental Predictors of Cognitive and Noncognitive Skills

The absence of racial gaps on personality assessments is striking b­ ecause ­children growing up in more privileged home settings tend to have more robust personality traits. The f­amily environment—­including exposure to poverty—­and parental education are significant predictors of positive personality traits. The NLSY 1997 researchers created a Family/Home Risk Index to mea­sure exposure to unfavorable living and ­family conditions. The index includes data on the following concepts:

88 • Chapter 3

• Cleanliness of home exterior and interior, as reported by interviewer • Neighborhood safety • Access to quiet space for study, dictionary, and computer • Religious engagement • ­Family routines • Parental disability • Parenting style (neither permissive, nor authoritarian), warmth, and level of monitoring Black ­people and Hispanics are strongly disadvantaged according to this index, whereas Jews, Mormons, and Asians are strongly advantaged. White Baptists score close to average but considerably below white Catholics. Roughly one full standard deviation separates the home environment of Jews from black ­people and roughly 0.60 standard deviations from non-­Jewish white p­ eople and black p­ eople. ­These conditions are strongly predictive of personality characteristics. ­Children raised in the least-­risky quintile (top 20 ­percent) on the f­ amily risk index score half a standard deviation better on the personality traits conducive to higher income and health than c­ hildren raised in the bottom quintile. This means the c­ hildren raised in a predictable, safe, and supportive home environment are more disciplined, enthusiastic, and less anxious. This translates into better health and higher incomes l­ ater in life. This same environmental variable also strongly predicts variation in cognitive ability and IQ. C ­ hildren raised in the top-­quintile environment score over one standard deviation higher on cognitive per­for­mance than c­ hildren raised in the higher-­risk home environment. Even when parents are equally educated, ­those who create a more supportive, safer, and warmer home environment raise c­ hildren who score higher on cognitive tests. None of the above evidence offers definitive estimates of the causal effect of parenting on cognitive per­for­mance or personality development, since ­children are not randomly assigned to families. Yet, such evidence does exist. Adoption studies have found very large effects on IQ when low-­income ­children are placed in the homes of affluent parents.28 A study in France found that adoption into a high socioeconomic status f­ amily—­with a man­ag­er or professional parent—­raised the IQ of c­ hildren aged four to six years old with low IQ by 1.3 standard deviations (19.5 IQ points) by the time they reached the age of 13, which is much larger than the U.S. black-­white IQ gap. ­Children

–0.69

–0.55

–0.36

–0.3

–0.26

–0.18

–0.09

0.09

H

ic an isp

0.147

Source: NLSY 1997. Means are calculated using 1997 sample weights.

Figure 3.2. ​Home environment risk index by race and ethnicity in standard deviation units with mean zero

, , , ic ic ic t, an lic e on e ish e cif er tis ite an ial an te o ric ian a t p p p w m h hi t t i e P s s r i i a t d c h e i i h J h o h d B w H ra H w or n Ca c w w Am In M cw n- ulti nic n Isla i ic i o o n a n i n n a N m N a a a As sp isp isp isp Hi H H H n n nnno no no no

–0.8

–0.6

–0.4

–0.2

0

0.2

0.4

ac Bl

k

0.33

90 • Chapter 3 0.47 0.18

0.08 0.05 –0.01 –0.13

–0.08 –0.28

–0.28 –0.61

Lowest-risk quintile

2 Personality traits

3

4

Highest-risk quintile

Cognitive ability

Figure 3.3. ​Cognitive ability and personality by riskiness of home and f­ amily environment Note: Expressed in standard deviations with mean of zero. Source: NLSY 1997. Means are calculated using 1997 sample weights.

a­ dopted into a working-­class f­ amily also benefited, but by much less (0.5 standard deviations). Likewise, both IQ scores and personality mea­sures improve when ­children are randomly assigned to high-­quality preschools.29 ­These studies often find that the IQ effects do not last long a­ fter the programs end but the personality effects do; since personality is no less ge­ne­tic than IQ in standard twin estimates, one implication is that cognitive achievement requires sustained exposure to enriching environments during childhood, whereas long-­term benefits to personality may come from the confidence gained from even temporary cognitive benefits and the supportive environment that created t­ hose benefits.30 All of this evidence demonstrates that exposure to less risky f­ amily environments and access to high-­quality schooling would dramatically boost the earnings potential and health status of many black ­people, Hispanics, Native Americans, and working-­class white ­people. Moreover, even some ­children from rich families would be better off if their parents created warmer, more routine home environments. Differences in ­Actual Per­for­mance and Productivity

So far, the discussion of merit-­based pay has focused on individual traits that predict income and good health, such as educational attainment, cognitive

Merit-­ Based Egalitarianism • 91

ability, and self-­reported personality profiles. Even if one accepts that payment based on t­ hese traits alone would narrow the income distribution, one could argue that payment based on ­actual per­for­mance would not generate the same egalitarian result. For example, the founder and owner of a start-up business is compensated directly based on the per­for­mance of his or her com­pany. If profits grow ­because of successful innovation and expansion, the owner gains at least a share of t­ hose profits. Likewise, employees in many large corporations are paid according to ­actual mea­sures of per­for­mance, what­ever they might be.31 A body of psy­chol­ogy research investigates the f­ actors that predict better per­for­mance. Both personality traits—­especially conscientiousness—­and cognitive ability are strong predictors.32 Cognitive ability seems especially relevant for the per­for­mance of complex tasks, but less so for other types of work. Psychologist and Gallup se­nior scientist Frank Schmidt and co-­authors In-­ Sue Oh and Jonathan Shaffer recently analyzed 100 years of academic research on job per­for­mance and its relations to employee se­lection metrics.33 A number of fascinating findings comes out of this work First, job per­for­mance varies significantly but not nearly as much as the income distribution would suggest. The standard deviation in the value of output produced is roughly 40 ­percent of the average employee’s salary, so if the mean salary for an occupation is $100,000, then a worker in the 84th percentile produces $140,000 in value compared to $60,000 in value for a worker in the 16th percentile. The implication is that some in­equality in earnings could certainly be based on per­for­mance, but this in no way justifies the extreme disparities in payment observed in the United States, among CEOs for example. Second, cognitive ability (IQ essentially) is the best single predictor of job per­for­mance identified in the lit­er­a­ture for applicants to a new role. Its error-­adjusted correlation with job per­for­mance is 0.65  in the selection-­ science lit­er­a­ture (i.e., the body of research on how to predict employee per­for­mance). It works much better for professional and managerial roles (error-­adjusted correlation of 0.74) than for blue-­collar jobs (error-­adjusted correlation of 0.39). Yet, other mea­sures that have zero or even negative correlations with IQ are also highly predictive of job per­for­mance. ­These include mea­sures from tests of integrity (0.46), conscientiousness (0.22), and emotional stability (0.12).34 Assessments from structured interviews are only weakly correlated

92 • Chapter 3

with IQ but almost as predictive of job per­for­mance (.58 versus .65) and add 18 ­percent to the validity of the assessment when combined with IQ.35 Thus, a conscientious candidate who interviews well but has below-­average IQ is likely to outperform a candidate with high IQ whose interview is unimpressive and rates low on integrity or conscientiousness. A body of lit­er­a­ture in psy­chol­ogy has reported on race and gender gaps in per­for­mance evaluations—­both subjective and objective—of ­actual workers. The main finding from the most recent and comprehensive meta-­analysis from Patrick McKay and Michael McDaniel is that black-­white gaps in work-­ based evaluations exist but are small and constitute only about one-­quarter of the size of gaps in cognitive per­for­mance.36 Notably, the most recent meta-­ analysis from McKay and McDaniel finds substantially smaller black-­white gaps in per­for­mance than older studies, suggesting that the per­for­mance of black p­ eople has converged or that discriminatory effects of ratings have declined. Other scholars have found essentially no Hispanic-­white gaps in per­ for­mance across a smaller number of studies.37 The lit­er­a­ture includes vari­ous per­for­mance criteria, such as task per­for­mance, absenteeism, ser­vice complaints, commendations, and personality-­related ­factors. ­There are reasons to worry about racial bias in at least some of ­these mea­sures, but even with that caveat, the black-­white gap is small overall, and ­there is no gap at all on ­things like social skills and involvement in workplace accidents. ­A fter de­cades of research and refinement, Gallup’s management consulting scientists developed a survey instrument called the Q12, which mea­sures employee engagement. It asks workers to rate their level of agreement with 12 statements on a 1–5 scale. H ­ ere are some examples: • At work, I have the opportunity to do what I do best e­ very day. • ­There is someone at work who encourages my development. • At work, my opinions seem to count. • I know what is expected of me at work. • In the last seven days, I have received recognition or praise for d­ oing good work. When combined, the answers to ­these and the other seven items have extraordinary predictive power when it comes to per­for­mance. If you rank work units—­individuals or small teams within organ­izations—by employee engagement using Q12, ­those units in the top quartile (top 25 ­percent) score

Merit-­ Based Egalitarianism • 93

dramatically better than ­those in the bottom quartile on absenteeism, turnover rates, shrinkage (the dollar amount of lost merchandise), personal safety incidents, and product or ser­vice quality defects. The more engaged business units also performed better in terms of profitability, productivity, sales, and customer satisfaction.38 ­These results come from an analy­sis of 1.8 million employees across 82,000 business units over 230 organ­izations spanning 73 countries. Th ­ ere is further evidence that increases in business unit engagement leads to better outcomes in the f­ uture, whereas better outcomes are only weakly predictive of an increase in engagement.39 To investigate racial and gender differences in engagement, I analyzed two Gallup surveys.40 One includes all 12 items from the Q12 but is ­limited to a small sample (3,500 U.S. working adults) collected from a random sample of respondents from the Gallup Panel and weighted to be nationally representative. The Gallup Panel consists of p­ eople who have previously taken Gallup surveys—­through random se­lection—­and said they would be willing to be re-­contacted for an online survey. The second dataset is from the GallupSharecare Well-Being Index and includes a much larger number of respondents (85,000 U.S. working adults), but it only includes the three Q12 items that tend to be most predictive of engagement (the first three listed above). In both databases, engagement is correlated with higher levels of educational attainment. In the larger database, p­ eople with gradu­ate degrees are the most engaged and show a 0.12 standard deviation difference from ­those with a high school diploma, controlling for age, race, and gender; p­ eople with bachelor’s degrees are only slightly more engaged than ­those with a high school diploma (.05 standard deviations). Despite t­ hese educational differences, the results show very small differences between races and between men and w ­ omen in both databases. In general, Hispanics, white p­ eople, and Asians had essentially equal engagement levels, whereas black p­ eople scored just one-­tenth of a standard deviation (0.11) lower than white p­ eople. In the panel, the differences between black ­people and other groups ­were driven by responses to a small number of items: Black ­people ­were significantly less likely to say they know what is expected of them at work, that they receive recognition or praise for good work, and that their opinions seem to count. In the larger database, black p­ eople w ­ ere also less likely to say that they have an opportunity to do their best work and that someone encourages their development. Overall, on both databases, the difference between black ­people and white ­people was equal to the difference

94 • Chapter 3

between men and ­women, with ­women expressing higher engagement scores by 0.10 of a standard deviation.41 If one classifies workers as engaged when their average answer to the 12 questions is a 4 or higher (on a 5-­point scale), then 36 ­percent of black workers are engaged compared to 40 ­percent of white workers, 37 ­percent of all male workers, and 42 ­percent of all female workers. Th ­ ese are obviously small gaps. Moreover, the gender engagement gap goes in the opposite direction from the gender pay gap—­despite ­there being a strong correlation between engagement and pay. In the larger database, I can also see how religion related to engagement. Again, the differences are very small. Controlling for demographic characteristics and w ­ hether the respondent’s com­pany is growing, Protestants and Catholics tend to score slightly higher than other groups. Protestants score 0.10 standard deviations higher on engagement than Jews, 0.09 standard deviations higher than Mormons, 0.10 higher than ­those with no religion, and 0.15 higher than ­those who report other non-­Christian religions. Differences with Muslims are insignificant.42 Although t­ hese are not direct mea­sures of per­for­mance, their strong connection with per­for­mance and the quality of work suggests small differences in a­ ctual work per­for­mance between groups.43 This reinforces my claim that payment based on merit would not lead to significant differences between groups. ­There is some evidence to suggest that good management practices may be enough to eliminate the small racial gaps in per­for­mance. Recent evidence from one large com­pany shows that establishment-­specific gaps in black-­white per­for­mance can be entirely explained by the climate of the store. This result comes from an impressive study by Patrick McKay, Derek Avery, and Mark Morris, who obtained data from 6,130 workers across 743 retail stores.44 They found no black-­white differences in sales per­for­mance but did find that relative per­for­mance by store varied according to how employees rated the climate of the store on four questions using a 1–5 scale, where 5 means strongly agree. 1. I trust the com­pany to treat me fairly. 2. Top leaders demonstrate a vis­i­ble commitment to diversity. 3. The com­pany maintains a diversity-friendly work environment. 4. The com­pany re­spects the views of ­people like me.

Merit-­ Based Egalitarianism • 95 58.2%

Hispanic

58.0%

57.0%

White Other racial group

54.8%

54.4%

55.1%

Asian

Black

Men

60.0%

Women

62.6% 59.8%

59.5%

58.4% 56.9%

56.8%

56.5% 55.1%

M or m on No re sp on se No re At ligi he on ria / no n n– Ch O ris the tri r an

Je w ish

Pr ot es ta nt Ot he rC hr ist Ro ian m an Ca th ol ic

M us lim

53.0%

Figure 3.4. ​Share of U.S. workers who are highly engaged at work by race, ethnicity, and gender, 2009–2015. Share of workers who are highly engaged by religious preference, among white respondents only Source: Gallup-Sharecare Well-Being Index. The sample consists of 84,749 U.S. adults surveyed in 2016. Respondents w ­ ere asked to rate their agreement on a 1–5 scale to questions about their primary job. High engagement means an average score of 4.3 or higher on each of the following three items, which are taken from Gallup’s longer Q12 battery: At work, I have the opportunity to do what I do best e­ very day; ­There is someone at work who encourages my development; At work, my opinions seem to count. To be considered highly engaged, respondents must have at least one answer that reflects strong agreement (5) and none below agreement (4).

The research team calculated sales per hour of work for each employee using electronic data maintained by the com­pany and tied to each employee identification number. Black sales workers performed equally as well as white ­people on average, but t­ here w ­ ere impor­tant differences across stores. Black ­people performed worse in stores with a poor climate rating and

96 • Chapter 3

better in stores with a favorable climate rating. It stands to reason that residual black-­white gaps in per­for­mance could be reduced more broadly by promoting respectful workplace cultures. Entrepreneurship

Another impor­tant mea­sure of a person’s productivity capacity and per­for­ mance relates to her or his inclination and ability to start and run a successful business. Entrepreneurial activity is impor­tant for maintaining or increasing living standards b­ ecause unchallenged incumbent businesses would have ­little incentive to launch new and better products or lower prices with re­spect to quality ­unless they ­were challenged by potential or ­actual competitors. Likewise, some entrepreneurs launch entirely new and impor­tant goods or ser­vices in their own right. In the United States, ­there are very large gaps by race and gender in business owner­ship rates. In 2014, black ­people and Hispanics comprised 29 ­percent of the U.S. population but only 7.2 ­percent of business ­owners. The under­representation of Hispanics and black p­ eople is even worse in prominent venture capital funds and at tech companies.45 ­Women are also heavi­ly underrepresented. Yet, ­there is compelling evidence that this is the result of differences in economic opportunity, not entrepreneurial talent. Gallup scientists developed a “Builder Profile” to identify the p­ eople most likely to start and succeed as entrepreneurs, based on their personality characteristics and thought processes—­including their confidence, tolerance for risk, creativity, and other ­factors.46 The index predicts business per­for­mance in several ways—­whether the owner meets revenues and profit goals, the number of employees, the level of revenue, and revenue growth. P ­ eople with higher builder profile scores also live in richer zip codes. Yet, Gallup data reveal no racial or gender gaps in entrepreneurial potential. Black p­ eople and non-­Hispanics score no differently than white p­ eople. White business ­owners score significantly higher than nonowners, but so do black and Hispanic business ­owners and ­there are no racial differences between them. Among non–­business ­owners, ­there are also no differences between non-­Hispanic white ­people and Hispanics and black ­people. The implication is that an equal opportunity society would generate egalitarian rates of entrepreneurship between racial and ethnic groups and even similar success rates among business ­owners.47

Merit-­ Based Egalitarianism • 97

In sum, the research described above provides even stronger evidence that merit-­based pay would reduce income in­equality overall and between groups, even in the absence of efforts to make education and ­family environment opportunities more egalitarian. When combined with efforts to enhance the equality of opportunity, merit-­based pay would most likely eliminate race and ethnic gaps in income and health status.

BEYOND SKILLS

The chief prob­lem with ­labor market in­equality is not the market per se. Rather, it is the way po­liti­cal power operates to control and affect markets. This power is sometimes acute, such as when Wall Street banks and shareholders are “bailed out” by taxpayers, saving them from bad decisions and irresponsible be­hav­ior. Yet, more often, the power that shapes markets works gradually and cumulatively through de­cades of lobbying, cultural persuasion, and minor changes ­here and ­there to state licensing laws, tax laws, or other means of affecting how much ­people are paid. The notion that skills—­both cognitive and noncognitive—­determine pay is the basis for the “­human capital” theory. It has a lot of theoretical strength, as the above discussion illustrates, but it is also ­limited. The prob­lem, for the theory, is that social rules are a key determinant of pay for many occupations. This explains why t­ here is such variation in pay across industries and occupations, even ­after taking into account cognitive and noncognitive skills. ­Later chapters ­will discuss specific rules that shape some of ­these differences in pay across professional occupations. The importance of nonmarket ­factors is fairly obvious when considering compensation for public employees, which depends on the willingness of state and federal legislatures—­and their constituents—to pay for the ser­vices provided, including policing, fire prevention, education, and, in many countries, healthcare. The vagaries of politics, not productivity, explain why teachers are paid dif­fer­ent salaries in dif­fer­ent cities, a­ fter adjusting for the local costs of living.48 Likewise, t­ here is no reason to expect the “market” to determine prices in which the government is one of the largest buyers. That includes healthcare. It is po­liti­cal power—­not the market—­that leads to nurse prac­ti­tion­ers being paid only half as much as f­ amily physicians, even though they are qualified to perform the same ser­vices.

98 • Chapter 3

The bottom line is that if po­liti­cal equality prevailed, markets would produce far more fair and egalitarian outcomes. The traits that predict per­ for­mance and per­for­mance itself vary far less than ­actual income, and ­under conditions approximating equal opportunity, ­these differences would shrink further.

4 The Importance of Equal Access to Public Goods and Markets

In the spirit of Rawls and Plato, I define a just society as one that grants and defends the basic liberties of individuals, while it maximizes the welfare of the least advantaged members ­under conditions of po­liti­cal equality and provides equitable opportunities for every­one to do the work that naturally suits them. By basic liberties I mean the right of individuals to engage in mutually beneficial transactions with other parties and live as they see fit, provided their actions injure no one e­ lse. Th ­ ese stipulations follow from a basic princi­ple of reciprocal fairness. By the least advantaged, I mean something like t­ hose ­people who are at the bottom of vari­ous overlapping distributions for the good t­ hings in life: emotional and loving ­family and friendship networks; psychological and physical good health; the capacity to learn, understand, and influence the world; purchasing power; a sense of accomplishment; and re­spect from ­others. In princi­ple, the least well off would score in the bottom quintile on an index that mea­sures all of ­these ­things, if one existed. Maximizing their welfare ensures that every­one’s welfare is at least as high. By po­liti­cal equality, I mean several t­ hings: 1. Citizenship is available to all born in the territory and open to foreign-­ born immigrants u­ nder certain conditions, which I outline l­ ater in this chapter. 2. Each citizen holds equal power in choosing his or her leaders and equal protection ­under the law. Access to power is not based on hereditary ties, gender, tribe, ethnicity, race, nationality, or parental social status.

100 • Chapter 4

3. Public goods are provided to all through public funding and on an equitable basis. The most impor­tant of ­these are security, education, and the infrastructure of commerce and production (e.g., roads, bridges, ports, environmental resources, telecommunications). a. The education curriculum is universal and encourages youth at an appropriate age to specialize in and master occupations that suit their interests and, presumably, their natu­ral ability.1 b. Security applies to physical protection (e.g., police and military), protection of property (e.g., firefighting ser­vices), protection from injury or disease through healthcare, and relief from poverty, job displacement, and disability. 4. Each citizen is entitled to the basic liberties, including equal access to markets and the right to engage in mutually beneficial exchange as a customer or seller.

­F REE EXCHANGE AND PO­L ITI­C AL EQUALITY

Among the basic liberties, I follow Rawls in including ­those enumerated in most modern republican constitutions, including that of the U.S. Bill of Rights. However, I would also include one that is rarely mentioned, at least outside of libertarian philosophy: the right to engage in mutually beneficial exchange with o­ thers. The right to f­ ree exchange is as impor­tant as other fundamental liberties, such as freedom of expression, due pro­cess, and equality ­under the law. Without it, governments can block vitally impor­tant economic activity, thereby suppressing individuals or groups of p­ eople and preventing them from achieving health and po­liti­cal influence. In medieval Eu­rope, commerce was dominated by politics at the expense of ­free exchange. The largest economic organ­izations had to win f­ avor and mono­poly status from the ruler.2 At a more basic level, and therefore more significant for individuals, commercial activity within and across cities was dominated by guilds. As economic historian Sheilagh Ogilvie shows, guilds suppressed competition both by limiting rights to trade or transport products to members and by limiting membership to a thin slice of the male population, often ­those with hereditary privileges of owner­ship and deep community ties, as well as the right ethnic affiliation, who qualified as citizens.3

Access to Public Goods and Markets   •  101

­ ese practices eviscerated opportunities for upward mobility and directly Th lowered the living standards of most medieval Eu­ro­pe­ans. From the end of slavery in 1865 ­until the mid-­twentieth c­ entury, African Americans w ­ ere forced to live in a society that forbade them access to markets and to engaging in mutually beneficial exchange with o­ thers in fundamental ways. In the l­ abor market, African Americans could not become professionals in many states ­because state licensing boards and professional associations would not allow it. ­Until the late 1960s, the American Medical Association’s state and local affiliates frequently prohibited African American doctors from membership, depriving them of licenses, hospital affiliations, patients, and income.4 The American Bar Association (ABA) was even worse in that even the national organ­ization—­not just local or state affiliates—­ officially barred black would-be l­ awyers u­ ntil 1943, and the first black member was not admitted u­ ntil 1950.5 This was not for want of candidates. In 1911, the ABA’s executive committee formally rescinded the membership of William H. Lewis, who was born to newly freed slaves ­after the Civil War and was appointed assistant attorney general of the United States by President Howard Taft.6 The ABA stated that it was settled practice to admit only white men. Education in general has also commonly been distorted to disfavor po­liti­ cally oppressed groups. Most northern colleges and almost all southern colleges—­except t­ hose deemed historically black—­refused to admit African Americans well into the twentieth c­ entury. Despite its location in New Jersey—­not a former confederate state—­Princeton University did not admit its first black student ­until 1935.7 Similarly, access to housing markets through mortgage financing was also systematically stifled for African Americans. African Americans w ­ ere forced—in many cases by formal law and, in the rare cases when that failed, informal vio­lence—to live in segregated communities throughout the United States. Deed restrictions that forbade sales to African Americans, zoning laws, and real estate industry policies all prohibited black ­people from buying or renting homes in desirable neighborhoods, which, in turn, l­ imited access to desirable schools and neighborhood amenities.8 Segregated black communities, moreover, ­were deprived of ­g reat wealth and the benefits of investment and home owner­ship. In the 1930s, the federal government created tiered neighborhood-­based evaluations of credit worthiness, which w ­ ere quickly a­ dopted by private lenders and the federal

102 • Chapter 4

housing agencies that subsidized and secured mortgages. ­These standards systemically deemed black neighborhoods or even white neighborhoods with black population growth as unworthy of credit, thereby locking out a huge portion of the black population from access to financial markets.9 ­These standards prevailed during the massive post–­World War II housing boom, which other­wise dramatically improved living standards for many white working-­ class families.

­F REE MARKETS AND THEIR LIMITATIONS

Any full accounting of liberties must consider access to markets to be fundamental. By equal access to market, I mean that markets are open or “­free” in the sense that all exchanges are permitted that are voluntary and mutually beneficial between parties, with some impor­tant caveats and exceptions. Exchanges that are coercive or made u­ nder fraudulent or false pretenses are prohibited. Hence, basic regulations on the quality of goods and ser­vices w ­ ill likely be required for many products, and in some cases, this would justify licensing requirements, clinical t­ rials, or inspection from public officials. The same rule also applies to exchanges between workers and employers, which should be entirely voluntary, recognized as mutually desirable, and never conducted in an exploitative or misleading fashion. Likewise, exchange that harms the public—­such as exchange that results in pollution—­would also be subject to regulation, such as a pollution tax or prohibition, depending on the seriousness of the harm. Such exchanges violate the princi­ple that the exercise of basic liberties should injure no one. Exchanges between citizens and noncitizens are permitted ­under the same conditions, resulting in f­ ree trade of goods and ser­vices. In this context, it should be clear that I do not mean “­free trade” to refer to the absence of tariffs on imports. As I mean it, ­free trade would not permit exchanges for goods produced by slave laborers or workers who are coerced into oppressive working conditions. In practice, p­ eople with dif­fer­ent po­liti­cal views and standards may disagree about ­whether a specific factory or workplace should be designated as oppressive, but in princi­ple, such disputes could be effectively resolved by in­de­pen­dent third-­party agencies, nonprofit organ­izations, and courts. In practice, at least since Karl Marx, po­liti­cal pressure from activists and student groups has often been able to hold corporations accountable for reforming

Access to Public Goods and Markets   •  103

and improving conditions in factories, thus making trade more mutually beneficial than it would other­wise be. Assuming the exchanges qualify as mutually beneficial as described above, prohibitions and even tariffs on imports should generally be regarded as unjust when they go beyond minimal sales taxes on imports, which may be needed for government revenue, to penalties meant to deter the consumption of foreign products or protect domestic companies from competition. The reason restrictions on trade are generally unjust is that they prevent mutually beneficial transactions. If you, as an individual or business owner, can get a better deal (i.e., a higher quality-­to-­price ratio for the product), you should be able to make that deal, so long as it other­wise meets the conditions laid out. When buyers have access to the best available deal, welfare increases, as the quality of a product increases relative to its price. Such transactions also have dynamic effects in that they increase the number of competitors in a market, encouraging innovation and limiting profit margins. In this way, open markets do not benefit only the two parties who are exchanging goods; they also create broader benefits to all t­ hose who participate in the market as buyers—­either directly (as customers) or indirectly (as buyers of products that use imports as components of a final product). One argument against f­ ree trade is that the beneficiaries—­even if large in number—­should not be allowed to benefit at the expense of even a small number of local producers (e.g., factory ­owners) and their employees. My response is that the right to trade is a basic liberty, but ­there is no meaningful right to work in a specific industry, occupation, or com­pany, nor could t­ here be in any practical sense. ­There are countless reasons why a layoff may occur or a firm close down, including poor managerial choices, poor product quality, supply-­chain prob­lems, and competition from domestic companies. ­There is no ethical obligation for taxpayers to bail out a slumping clothing store or restaurant so that its employees do not have to seek out new work. Th ­ ere is, however, an impor­tant consideration in determining how producers are harmed by foreign imports. The issue is ­whether harm comes from cheating or fair competition. If it is through cheating—­such as corporate espionage, intellectual property violations, or forced transfers of technology as a condition to market participation—­then the domestic producers have a legitimate complaint, and the foreign producers should not be allowed to participate in the market. The National Bureau of Asian Research and scholars such as Robert Atkinson, the president of the Information Technology and Innovation

104 • Chapter 4

Foundation, have documented that Chinese companies routinely commit ­these abuses at the expense of U.S. and other companies.10 ­These practices do not meet any au­then­tic definition of ­free trade and should be rejected, even if short-­term economic costs are large. At the same time, it must also be recognized that Chinese entrepreneurs have made genuine improvements in efficiency in many industries.11 Chinese firms that invest in R&D and win intellectual property protections are much more likely to export than t­ hose that do not. This does not prove that the gains in Chinese market share are the result of fair competition, but it does suggest that successful Chinese companies are taking actions that create social value.12 If ­there is agreement on the enforcement of rules against cheating, the key question is: Should companies that offer products at a lower quality-­to-­price ratio be compensated when a domestic, or even a foreign, competitor beats them fairly? The answer must be no; other­wise anyone could open an inefficient and poorly managed business and cash in on incompetence. Reciprocity does not imply that ­those who lose ­under fair competition should be sheltered or compensated. In the context of a welfare state, with protections against poverty, entrepreneurs must be allowed to risk and experience failure. They also must be allowed to realize gains when they succeed; if their superior products ­will be taxed u­ ntil the quality-­price ratio is dragged down to their competitors’ level, the entrepreneur’s motivation to make sacrifices, take risks, and invest ­great effort into starting and growing a business ­will vanish, making many p­ eople worse off and depriving the market of innovation. Thus, tariffs cannot be justified as a means to protect domestic producers. The logic of this position is clear when applied outside of international relationships to other economic relationships. If tariffs, taxes, or other restrictions on exchange made with the aim of sheltering lower-­performing businesses could be justified, then local governments or state governments would have e­ very right to protect their retail establishments, restaurants, theaters, entertainment complexes, hospitals, and other businesses from competitors in neighboring states or even local jurisdictions. Should Washington, DC, residents pay a 20 ­percent tax to protect hospitals or surgeons who work within the District of Columbia if they see a specialist surgeon at Johns Hopkins University in Baltimore? Should a cure for cancer that is developed by a phar­ma­ceu­ti­cal com­pany in Minnesota be taxed by the New Jersey legislature u­ ntil the price in cost per quality-­adjusted life year is the same or higher than a New Jersey–­based phar­ma­ceu­ti­cal com­pany?

Access to Public Goods and Markets   •  105

Such interventions would rightly be regarded as corruptions of the po­liti­ cal pro­cess and eco­nom­ically self-­defeating from a national perspective. The same logic applies to international relationships. Some may argue that restrictions on international (or domestic) trade could be justified by concerns for the worst off, if one counts laid-­off workers—or ­those in their communities—as being among the most vulnerable members of society. ­There are two reasons why this logic does not justify tariffs. First, neither manufacturing workers nor former manufacturing workers are among the worst off on any obvious mea­sure. Manufacturing and even export-­ oriented ser­vice jobs tend to pay well, but available evidence suggests that workers laid off in manufacturing experience smaller wage decreases relative to workers in other sectors.13 Manufacturing workers also enjoy higher levels of job security relative to workers in domestic-­facing sectors, despite competition from China and other countries. Comprehensive government data show that the risk of layoff is considerably higher for the average nonmanufacturing worker relative to a manufacturing worker, especially construction, retail, and even professional ser­vices.14 Laid-­off workers from any sector deserve sympathy and compassion, and a republican state should create and maintain supportive programs to smooth the transitions to new jobs, and if needed, new c­ areer paths. It ­shouldn’t ­matter if the source of competition that led to the layoff is foreign or domestic. A second reason that restrictions on domestic or foreign trade cannot be justified as a means for protecting the worst off is that by protecting some workers, they harm ­others—­and almost always a much larger number. Lower-­ income workers spend a larger percentage of their incomes than richer workers, b­ ecause it is harder to afford savings when you have less money. Raising the price of t­ hings they buy is the same as lowering their purchasing power. Protecting a small number of manufacturing workers through tariffs takes money from a massive group of poor ­people—­many of whom ­don’t even have jobs—­and gives it to p­ eople in manufacturing who do have relatively decent paying jobs and, since they are in the ­labor force, are unlikely to be suffering from ill health or disability. A final justification for restrictions on trade is to make domestic companies stronger. It is dubious as to ­whether protecting businesses from competition actually makes them stronger. A comprehensive investigation by the World Bank of East Asian “infant industry” practices found that innovation—­ measured via productivity growth—­was slowest in the most protected

106 • Chapter 4

industries and strongest for companies that faced strong international competition.15 Even if tariffs did make domestic companies stronger, they would be unfair to both foreign producers and domestic businesses and consumers who would prefer to buy the foreign product at its market price, rather than a heavi­ly taxed price. In this framework, tariffs to support infant industries are no dif­fer­ent from any other form of “corporate welfare” designed to tax consumers or businesses in order to support special interests. Such tariff policies should be rejected, regardless of their “strategic” benefits. This does not mean that supporting business or “competitiveness” is fundamentally problematic. ­There are many legitimate ways governments can make their domestic businesses more valuable: workforce development, subsidizing R&D, creating high-­functioning national laboratories and encouraging business spin-offs, enforcing smart intellectual property laws, and ensuring a robust venture capital market, as well as through direct government grants. The economist Mariana Mazzucato has laid out best-­practice princi­ples and policies in this area, as has Robert Atkinson of the Information Technology and Innovation Foundation.16

IMMIGRATION AND THE RIGHTS OF NONCITIZENS

Immigration is generally highly beneficial for the receiving country ­because population density fosters occupational specialization and greater productivity, as Plato, Adam Smith, and con­temporary urban economists all agree. In their roles as workers, entrepreneurs, and consumers, immigrants contribute to the affluence of society. ­There are two impor­tant caveats. Immigration affects the distribution of income by redistributing from the groups who compete most closely with the immigrants in the l­abor market (or for public resources) to the groups who employ immigrant workers or work with them as employees or co-­workers. Second, while immigrants benefit society as workers, they, like every­one ­else, absorb public resources such as antipoverty assistance, health, retirement, and other welfare payments. When immigrants disproportionately work in the lowest-­paying occupations, their net effect on public resources is considerably greater than that of t­ hose who work in the highest-paying occupations. In a recent book, the Harvard economist and immigration expert George Borjas makes ­these points with unusual clarity and care. He is one of the

Access to Public Goods and Markets   •  107

world’s leading scholars on immigration and has published dozens of articles in top peer-­reviewed journals.17 Borjas is clear that if the only national goal ­were to maximize income, immigration policy would allow virtually any number of highly educated immigrants in but no “low-­skilled” immigrants—­ meaning they lack a college degree. A massive influx of highly skilled immigrants would redistribute income from the affluent to the less affluent, boost economic growth, and lower the tax burden on native-­born Americans. Despite the seductive logic of favoring only highly educated immigrants, Borjas rejects the idea that immigration policy should be closed to t­ hose with low levels of education, and I agree. Th ­ ere are impor­tant moral and civic reasons to accept mi­grants, even when their economic contribution is predicted to be modest. Allowing less-­educated workers entry and integration into society provides economic and social mobility for the workers and their descendants, thereby creating a slightly more prosperous world and elevating the disadvantaged. This pro­cess occurred for Eu­ro­pean and Asian immigrants to the United States, who w ­ ere often poor and unskilled upon arrival, but in subsequent de­cades converged or even surpassed the Anglo-­Saxon native population with re­spect to average income and education levels, as I discuss in chapter  6. Recent Asian and African immigrants have already converged, and Hispanics ­will soon reach parity with non-­Hispanic white p­ eople, if current trends continue.18 In this way, it would be an empirical error to categorize a group of immigrants as permanently “low-­skilled.” A nontrivial number of less-­educated mi­grants ­will also make large and direct contributions to their new society by starting valuable enterprises, assisting in the generation of innovative practical and cultural ideas, and becoming caretakers, security providers, partners, friends, companions, or spouses to the native-­born population. It is impossible to precisely mea­sure let alone predict the value that might come through t­ hese sorts of activities and relationships. The cultural exchange that mi­grants carry as they come and go across borders also ensures more harmonious relations between countries and allows good ideas and innovations to diffuse more widely and quickly than they other­wise would. When mi­grants are a small proportion of the native workforce, the effects on ­labor markets w ­ ill also be small, but mass immigration raises more troubling issues that warrant careful consideration, as I describe next. With re­spect to the worst off, immigration policy differs from ­free trade in that it puts the poor at greater risk. The four lowest-­paid major occupational

108 • Chapter 4

categories in the United States for workers aged 36 to 45 are food preparation and related occupations; personal care and ser­vice occupations; building and grounds cleaning and maintenance; and farming, fishing, and forestry. It prob­ably ­isn’t a coincidence that ­these occupations are heavi­ly staffed by foreign workers: 23 ­percent of all workers aged 36 to 45 are foreign born, but they make up 39 ­percent of workers in ­those four low-­paying occupations. To investigate this further, I examined income data for 1.6 million workers born in the United States and calculated the relationship between incomes from 1994 to 2017 and the percentage of workers in their occupational field who are foreign born. I found that for e­ very 10-­percentage-­point increase in the foreign-­born share of workers in an occupation, native-­born income was 2  ­percent less. If an occupation went from 10  ­percent foreign born to 60 ­percent foreign born in a given state, the model predicts native workers would see a 10 ­percent reduction in income. This effect is a bigger prob­lem for less-­educated workers ­because they are in occupations with higher rates of foreign-­born workers and, as a result, are more likely to experience an influx in competition.19 ­These correlations do not necessarily indicate that immigration depressed wages. Few topics are more controversial in economic research, but a review of many economic studies on the subject does suggest that workers with low levels of education often see small but nontrivial reductions in wages as a result of the immigration of workers with similar skills.20 ­These income reductions would be only partially offset by increased purchasing power that resulted from entrepreneurship and lower prices for ser­vices. My general conclusion a­ fter reviewing the data and academic lit­er­a­ture is that large-­scale immigration of workers with low levels of education has exacerbated income in­equality by hindering growth below the median of the income distribution. Immigration has not made any impor­tant difference to the rise of the top one p­ ercent’s share of income, as I mentioned in chapter  1, but it still raises impor­tant issues for how to or­ga­nize a just society. Likewise, immigrant l­ abor must be regulated differently than trade b­ ecause of how immigration interacts with citizenship and po­liti­cal power. In just socie­ties, citizens are required to have equal access to public goods from birth to death and to power in adulthood. The education of citizens prepares them to maximize their productive capacity and freely choose the occupation that suits their ability. This arrangement is needed to meet reciprocal obligations to one another, but it also has the practical benefit of sustaining social and

Access to Public Goods and Markets   •  109

physical security systems, including antipoverty relief; a society with low labor-­ force participation or a high share of low-­paid workers with unrealized potential ­will have difficulty meetings its obligations. A policy of open borders would thus risk collapsing the po­liti­cal and economic foundations of a just society. For t­ hese reasons—­political stability and the protection of the worst off— it is ethically impor­tant to impose limits on the influx of immigrant workers in order to fulfill obligations to fellow citizens, despite the fact that immigrant workers—­even t­ hose with l­ ittle formal education—­contribute to economic growth and lower prices for consumers. What is the greatest good for the greatest number sometimes conflicts with the greatest good for the worst off. A related ethical question pertains to ­whether or not we should give preference to fellow citizens over noncitizens, other ­things being equal. I believe the answer is yes due to reciprocal moral obligations. Living in the center of the vast cosmopolitan Roman Empire, Cicero nevertheless gave priority to local relationships. He writes: “It is still a more intimate bond to belong to the same city; for the inhabitants of a city have in common among themselves forum, ­temples, public walks, streets, laws, rights, courts, mode and places of voting, beside companionship and intimacies, engagements and contracts, of many with many.”21 In other words, the sharing of public goods, laws, and commitments, as well as commercial exchange, all create tighter bonds than between strangers, even between kin living in dif­fer­ent cities. Likewise, security considerations speak to the importance of shared citizenship bonds. It is one’s fellow citizens who fight in times of war to protect one’s property and life or engage in protests to win rights or protect liberties, and even if t­ hese w ­ ere the only bonds, it would warrant preferential treatment from the perspective of reciprocal altruism. While Cicero valued shared local bonds, he was not indifferent to noncitizens or foreigners. He praises past Romans for generosity in war and allowing former e­ nemy combatants to become citizens when they fought honorably. He also develops a compelling princi­ple: “What­ever one can give without suffering detriment should be given even to an entire stranger,” even as he emphasizes the limit is a necessary one “so that we may have the means of generosity to ­those peculiarly our own.”22 He finds p­ eople at fault who give to strangers “what o­ ught to be employed for the needs of their own families or bequeathed for their ­future use.”23

110 • Chapter 4

The implication for immigration policy is that foreigners should be freely admitted up ­until the point when their participation in ­labor markets begins to seriously harm the least advantaged domestic workers or when their requirements for aid—if many are disabled or too old to work, for example—­ impose a harmful burden on the nation. In recent de­cades, I believe the United States may have exceeded this threshold. The foreign-­born share of the U.S. population spiked from 4.7 ­percent in 1970 to 13.7 ­percent in 2017. The current figure is near the long-­term average from 1850 to 1930, when immigration fueled the growth of a new country.24 As other chapters describe, racial and ethnic conflict was extreme during that time, and economic historians conclude that mass immigration contributed to income in­equality by lowering wages for blue-­collar workers.25 Since 1970, the annual net flow of immigrants has been roughly 743,000, which is 20 ­percent as many as the total number of annual births. If annual net flows w ­ ere reduced to 400,000, the foreign-­born share of population would stabilize at around 10 ­percent.26 Immigrants to the United States hold ­every level of education and skill but have been heavi­ly represented among the poorly educated: 27.5 ­percent of the foreign-­born population aged 25 and over has not completed high school compared to only 8.7 ­percent of U.S. natives.27 Reducing the net inflow from this group—­many of whom are not authorized—­would go a long way to reducing harm to the U.S. economy and the disadvantages to current U.S. residents. Reducing the net flow of mi­grants should not be considered a border security prob­lem. In fact, the best available evidence on the effects of border security is that it has been counterproductive and exacerbated rather than reduced the negative consequences of immigration. Prince­ton sociologist Douglas Massey and his collaborators have been surveying Mexicans and Mexican residents living in the United States ­every year since 1982, collecting detailed information about migration and border crossings. The startling conclusion from this research is that border security has locked mi­grants in the United States, preventing the natu­ral back-­and-­forth flow that other­wise would have taken place. From 1970 u­ ntil the mid-1980s, more than half of Mexican immigrants—­both t­ hose with l­egal documents and t­ hose without them—­returned to Mexico within a year of working in the United States, but investments in border security caused the probability of returning to plummet to 20 ­percent in 2007 for undocumented workers, even as 92 ­percent of

Access to Public Goods and Markets   •  111

documented mi­grants returned to Mexico within a year.28 Massey, writing with Jorge Durand and Karen Pren, finds that border security has no effect on the likelihood of migrating from Mexico to the United States, only on the likelihood of leaving the United States to return to Mexico. As a result, restricting ­legal work visas and increasing the risk of crossing the border did not reduce the number of undocumented workers in the United States; it increased it. The evidence strongly suggests that legalizing migrations from Mexico and easing border crossings would reduce the number of permanent Mexican-­born residents in the United States. Border security and mass deportations are completely unnecessary. Paradoxically, empowering immigrants po­liti­cally by granting them ­legal work and residential status ­will reduce the net flow of mi­grants dramatically. While mass migration can harm the poor, ­there are several other impor­ tant caveats to be kept in mind when considering the justice of immigration policy. One is that rich countries can afford to take on thousands of even the poorest refugees, and d­ oing so, bring benefits that go beyond ­simple economic considerations. The harm imposed on noncitizens by prohibiting their entry should be compared to the harm imposed on the least well-­off citizens by allowing their entry. Noncitizen mi­grants fleeing extreme po­liti­ cal persecution and high risk of death should arguably be given priority over the threat of small wage reductions, even for disadvantaged members of society. Another point more closely tied to economics is that ­there ­will be job vacancies that simply cannot be efficiently filled using domestic ­labor ­because of local shortages or the undesirability of the work in relation to other opportunities. Thus, visas for less-­educated immigrants should be inversely proportional to mea­sures such as the local unemployment rate for workers in that occupational category and proportional to hiring difficulty or income growth. Such policies increase the chances that immigrants w ­ ill complement rather than compete with the worst-­off domestic workers. In recent years, jobs that require specialized technical skills are t­ hose most likely to show signs of hiring difficulty and also trigger a disproportionate number of visa requests from employers.29 For ­those who do enter, the more empowered immigrants are to pursue the occupation that best suits their abilities, the more positive w ­ ill be their effect on society. A practical implication is that workers should not be closely tied to any one employer for ­legal status, so that they can pursue employment

112 • Chapter 4

relationships that best use their talents. Likewise, t­ hose who enter with temporary work visas should be readily able to convert them to permanent work visas on a pathway to citizenship, if they choose to follow it, and the children of mi­g rants should be given full access to public ser­vices when they qualify. To summarize, equal access to markets in the context of immigration policy means two ­things. First, it means regulating and stabilizing the flow of mi­g rants so as to protect native citizens from wild distortions in the l­ abor market and preserve a close relationship between domestic residency and citizenship; part of ­those regulations should entail penalties to employers for violating ­labor laws. Second, it means granting l­ egal access to a large but proportionally manageable population of mi­g rants, so they can come and go freely, move up socially, gain access to public ser­vices, and be empowered to insist on mutually beneficial exchanges with employers. All authorized mi­ grants, u­ nder reasonable conditions and requirements, should be allowed to become citizens and encouraged to do so.

THE BASIC CONDITIONS AND INEQUALITIES IN SOCIAL STATUS

The above discussion is meant to clarify what I mean by access to public goods and markets in a just society through the lens of two especially controversial market topics—­namely, trade and immigration. This is not meant to suggest in any way that t­ hese are the only relevant issues. L ­ ater, I discuss in­equality in access to education (chapters 5 and 6), housing and neighborhoods (chapters 7 and 8), and professional ser­vices (chapter 9). Reforms spanning all of ­these spheres of economic in­equality, as well as ­those not mentioned, would be needed to meet the general conditions of a just society described above. If implemented, I believe t­ hese changes would radically reduce income in­equality, along with inequalities in subjective well-­being, education, health, and other ­things that m ­ atter to a good life. And yet, it is worth pointing out that such a society would not be a communist utopia with every­one possessing the same spending power and living standard. The vast gaps in welfare found in con­temporary rich countries such as the United States are in many ways the result of differences in po­liti­cal power. However, po­liti­cal equality would still result in some in­equality in income and welfare.

Access to Public Goods and Markets   •  113

Modest differences in welfare, or income specifically, are an inevitable (one could say natu­ral) quality of the world b­ ecause economic relationships are highly complex and dynamic, path dependent, and affected by random chance. Moreover, not every­one has the same preferences or aptitude for work. Modest wealth in­equality exists even in hunter-­gatherer socie­ties where occupational roles are virtually identical, members use the same technologies, and po­liti­cal equality prevails; anthropologists estimate wealth in­equality in such socie­ties to be comparable to income in­equality in modern Scandinavia, but roughly one-­half the level of income in­equality observed in the United States (and roughly one-­quarter of the level of U.S. wealth in­equality).30 I would argue that some market-­based in­equality is inherently consistent with basic notions of fairness and a reasonable conception of equality of opportunity. Th ­ ere are several ways that merit or skill differences would emerge even in a just society. One way is through innate differences, or their interaction with society. Individuals differ in terms of innate ability or attributes. I do not believe ­there are compelling moral grounds for refusing to reward t­ hose lucky enough to have certain ge­ne­tic advantages, including personality and physical attributes. Though their gifts may be unrelated to effort (and effort itself likely has some ge­ne­tic basis), it is entirely fair that more attractive models and actors should be paid more if they add greater value to a marketing campaign or artistic production, just as the best athletes should be paid more if their participation in a professional sports league attracts more viewers and fans, ­whether or not they had to train as hard as less-­gifted athletes. Likewise, a sales worker with a more charming personality should be rewarded if she or he is able to close sales deals more effectively than someone who comes across as less charming. It would likewise be foolish and immoral to eliminate any economic advantages that may accrue to inventors and entrepreneurs, w ­ hether or not their genes gave them an edge in sparking the creativity, motivation, or insight that ­others lacked. Differences would also emerge in compensation across occupational categories, ­because the ­labor market would play a role in setting the price of work. Yet, ­here a just society would look very dif­fer­ent from the con­temporary United States. In a just society, p­ eople would be educated in such a way that they would be capable of ­doing the work they found most valuable and fulfilling, which might oftentimes not be the work with the highest social status or income. The wages of ­people in many “low-­skilled” occupations are

114 • Chapter 4

presently far below ­those in professional occupations, but this is partly ­because in­equality in educational opportunities has pushed too many ­people into low-­skilled occupations. Th ­ ere would be more balance in a just society, such that the salaries of professionals would fall much closer in line with the salaries of h ­ ouse cleaners, day-­care workers, and hair stylists as a result of supply and demand. As imbalances arise, more ­people ­will move into that line of work u­ ntil balance is restored. This is not to say that all occupations would have the same average income levels in a just society. I think not. It would still be the case that some c­ areers would lend themselves to higher compensation ­because of higher levels of productivity, longer training requirements, the level of experience needed, or the demonstration of excellence at lower levels of work (i.e., man­ag­ers, who work their way up from entry-­level positions). C ­ areers that combine t­ hose advantages with the development of innovative new technologies and entrepreneurship would likely garner the greatest financial rewards in a just society. Individuals ­will also differ based on ­family circumstances, even in a just society, which can affect c­ areer outcomes and general well-being. Po­liti­cal equality and equal access to public goods could never eliminate e­ very relationship risk, such as the death of a parent or being born to a parent with psychological prob­lems. The best a just society could do, perhaps, would be to ensure that best-­practice ­mental and physical health treatments are available to families who need them and to other­wise shore up the conditions and capacity of the worst off with re­spect to bad luck or ge­ne­tic disadvantage. A more difficult case involves the transition from an unjust to a just society. In such cases, ­family inequalities may be quite large and based on historical injustices, such as slavery, colonization, or some other form of po­liti­cal oppression, and t­ here it would be appropriate for the government to purposefully elevate the conditions of the least well off. For example, I’ve already provided evidence that the average African American child grows up in a less privileged home environment; other chapters reinforce the point and describe the historic and current ­factors that produced and reproduce that disadvantage. Giving low-­income African American families extra resources—­such as health and ­family counseling, educational support, and tutoring for ­children—­would be a fair way to address this historical injustice without placing excessive burdens on the rest of society. ­These interventions could also be justified as a means to enhance the well-­being of the least well off.

Access to Public Goods and Markets   •  115

Even if ge­ne­tic and parenting conditions ­were identical—­say, for identical twins raised together by their biological parents—­inequalities could still emerge as a result of differences in the quality of training and education. It would be impossible to assign teachers of exactly the same quality to ­every child. Some teachers are better than o­ thers as a result of personality differences or other innate traits, or ­because they themselves had better training. What should m ­ atter is that the best teachers are not systematically assigned to ­children with greater po­liti­cal power, and society should take active steps to ensure that overall teaching quality is maintained at a high level. Parents should be ­free and encouraged to invest as much of their own resources—in terms of money or time—on private educational experiences that might give their child skills that distinguish him or her from peers. Distinguished excellence—­even if derived from private familial circumstances—is a benefit to o­ thers in a just society, whereas nepotism obviously is not. Still, a high-­ quality baseline education—­one that draws teachers and principals from a pool of highly competent professionals—­should be universally available starting as early as practical, such as age three. If the other criteria are met, any inequalities that arise u­ nder ­these conditions could be considered just, and in princi­ple they would meet Rawls’s criterion of benefiting the least well off, b­ ecause unequal compensation would be based on the productive contributions of individuals, rather than po­liti­ cal connections. Rewards based on productive contributions encourage innovation, investment, and productivity and lower the costs of goods and ser­vices, while enhancing their quality. The discussion above is far from comprehensive in terms of the complex set of issues that arises when considering the justice of a society, even at the most foundational level. Yet, it is intended to provide enough details to move forward with more practical considerations of how con­temporary democracies fail to meet the basic conditions sketched out ­here in several crucial ways.

5 Unequal Access to Education

This chapter starts by discussing how the United States has failed to deliver equal access to public education and how this failure creates in­ equality in the skills that are most predictive of income and health. It ends with a review of the latest scientific lit­er­a­ture on the ge­ne­tic basis of education and intelligence. The clear implication of this evidence is that environmental ­factors are the dominant explanation for why p­ eople vary in intelligence and education. Across individuals in the same birth year and growing up in the same country, 55 ­percent to 87 ­percent of variation in intelligence and education is explained by nonge­ne­tic f­actors, with twin studies showing higher ge­ne­tic effects than analyses based on ­actual genes. Across generations and countries, individual variation is almost entirely explained by nonge­ne­tic ­factors, and group-­level variation can be explained entirely by nonge­ne­tic ­factors. In­equality in educational opportunity has a long history in the United States. More than a de­cade before he escaped from slavery, Frederick Douglass, at the age of seven or eight, as far as he could remember, was transferred from a harsh plantation to serve a young ­couple, the Aulds, in Baltimore, who ­were related to his owner. Mrs. Auld, who had never owned a slave, soon began teaching him the alphabet and how to spell ­until, ­after a short time, her husband found out and forbade further instruction in no uncertain terms, telling her that literacy would spoil Frederick and make him unfit for slavery. Overhearing his master state that educated p­ eople could not be slaves, the young Douglass was emboldened to take ­every opportunity to further his education in the hope that it would prove a path out of slavery. Whenever he had a moment, often at ­great risk of retribution from his masters, he would study a newspaper or book. To further his instruction, he took bread from

Unequal Access to Education  •  117

the Aulds’ h ­ ouse and, while performing vari­ous ­house­hold errands, shared it with poor white boys who lived in his neighborhood in exchange for brief lessons. In this way, he taught himself to read well enough so that by age 12 Douglass could understand complex philosophical works. Eventually, he became internationally famous as one of the greatest orators, writers, editors, and public intellectuals in American history.1 Douglass’s example shows how extraordinarily difficult and unlikely it was for American slaves to become literate. The mandatory illiteracy of an oppressed group went hand in hand with po­liti­cal domination. Po­liti­cal in­ equality was the cause of their enforced illiteracy. Accordingly, the obverse is true. Po­liti­cal equality requires universal investment in and encouragement of the skills needed to thrive in a given society.

SCHOOLING AND THE DISTRIBUTION OF VALUABLE SKILLS IN A MARKET SOCIETY

When the U.S. Congress passed the 1964 Civil Rights Act, it included a provision that mandated a survey and report “concerning the lack of availability of equal education opportunities by reason of race, color, religion, or national origin in public educational institutions at all levels in the United States.”2 That study was led by a social scientist named James Coleman and became known as the “Coleman Report.”3 It was a massive undertaking, collecting data from over half a million U.S. students. The report confirmed perhaps not surprisingly that black students w ­ ere extremely segregated, even in the North, and that they attended schools with significantly fewer resources in the South. But some of its findings ­were surprising and unlikely to motivate policy change. ­A fter controlling for the race of students, Coleman found that “schools are remarkably similar in the way they related to the achievement of their pupils when the socioeconomic background of the student is taken into account. Differences in schools account for only a small fraction of differences in pupil achievement.”4 This finding—­that families m ­ atter more than schools—­has been echoed through the years by many liberal and conservative observers of education and in­equality in the United States, often citing Coleman’s work. For example, in his popu­lar but scholarly book Our Kids, which makes the case for investing in the success of poor ­children, the social scientist Robert Putnam argues

118 • Chapter 5

that opportunity gaps between ­children are largely the result of families, not schools: “The gap is created more by what happens to kids before they get to school, and by what kids bring (or ­don’t bring) with them to school—­some bring resources and ­others bring challenges—­than by what schools do to them.”5 Likewise, supporters of teachers’ u­ nions have used Coleman’s work to challenge the “No Excuses” idea, which is prominent in certain charter schools and asserts that all ­children should be expected to learn, regardless of their ­family circumstances. Instead, ­these critics of charter schools emphasize the importance of improving the conditions of students outside of school.6 Along ­these lines, education policy expert Diane Ravitch writes: “­Children who grow up in eco­nom­ically secure homes are more likely to arrive in school ready to learn than t­ hose who lack the basic necessities of life. If we are serious about closing the achievement gap, we should make sure that e­ very pregnant w ­ oman has good prenatal care and nutrition and that e­ very child has high-­quality early education.”7 It is certainly true that parents affect the level of scholastic achievement of their c­ hildren, and in a society with g­ reat in­equality in income and po­liti­ cal power, it should not be surprising that ­family life is also unequal. ­There are ways it can be made much less unequal—­via better-­functioning markets for housing and professional ser­vices, criminal justice reform, and public support for specific programs—­but it would be unfortunate if liberals concluded that t­ here is ­little that can be done through schools to reduce in­equality in cognitive per­for­mance of c­ hildren. Ironically, the view that schools are impotent in reducing cognitive in­equality aligns rather closely with the hereditarian school of thought, even though that perspective holds radically dif­fer­ent views about the fundamental origins of test score gaps. Hereditarians attribute social in­equality to ge­ne­tic differences. H ­ ere is what Richard Herrnstein and Charles Murray have to say about the Coleman report in their book, The Bell Curve: In affecting IQ, [schools] do not ­matter nearly as much as most ­people think. This conclusion was first and most famously reached by a study that was expected to demonstrate just the opposite. The report, issued in July 1966, announced that it had failed to find any benefit to the cognitive abilities of c­ hildren in public primary or secondary schools that could be credited to better school quality. The usual ways in which

Unequal Access to Education  •  119

schools tried to improve their effectiveness w ­ ere not likely to reduce the cognitive differences among individual ­children or ­those between ethnic groups. . . . ​The Coleman report did not prove that educational reform is always futile, but that, on the w ­ hole, Amer­i­ca had already achieved enough objective equalization in its schools by 1964 so that it was hard to pick up any effects of unequal school quality. The Coleman report tells us that the cognitive ability differences among individuals and groups alike on a national scale cannot be reduced much by further attempts to equalize the kinds of bricks-­and-­mortar ­factors and teacher credentials that school boards and taxpayers most often concern themselves with.8 Murray repeated this basic argument 18 years ­later in his book Coming Apart, which at its heart, argues that the dysfunction he sees in the white American working class is largely a result of their weak genes for intelligence and social competence. “Education does not have much of an effect on IQ ­after the child enters elementary school,” writes Murray. “The ­children of the well educated and affluent get most of the top scores b­ ecause they constitute most of the smartest kids. They are smart in large part b­ ecause their parents are smart.”9 ­A fter claiming that a child’s IQ score at age 6 ­will not be affected by age 18, even after an expensive private education, he writes, “This finding goes back to the famous Coleman report in the 1960s. . . . ​No one contends that education routinely transforms average c­ hildren into intellectually gifted ­children.”10 Hereditarians argue that genes largely determine test scores. Teachers’ ­unions argue that f­ amily environments largely determine test scores. Both groups are skeptical that the quality of schooling makes much of a difference. Coleman was a leading social scientist and a good one, but that does not mean he did the analy­sis correctly or used the best pos­si­ble methods, as the influential Stanford economist Caroline Hoxby has recently argued, and I draw on her argument in laying out my own concerns.11 The most impor­tant part of Coleman’s analy­sis compares test scores within the same school to average test scores across schools separately for each race. For black sixth graders, he finds that 20 ­percent of the variation in test scores can be accounted for by variation across schools, and for white p­ eople, schools account for somewhat less (14 ­percent) of the variation.12 Based on this, he concludes that school quality is a minor f­ actor in test scores.

120 • Chapter 5

But as Hoxby points outs, this evidence does not warrant Coleman’s conclusion and represents the biggest methodological flaw in his analy­sis. Unpacking the corollaries of test scores is not the same as identifying the causal ­factors. Moreover, Coleman’s method aims to explain variation among black ­people separately from variation among white p­ eople and other students. Using Coleman’s method, which analyzes races separately, it is pos­si­ble that 100 ­percent of the gap between black ­people and white ­people is between schools, whereas 0 ­percent of the gap between high-­scoring black ­people and low-­scoring black p­ eople is between schools. Coleman himself concedes the point: Why are the racial and ethnic groups separated in the analy­sis? Let us suppose that all Negroes go to equally bad schools and all whites go to equally good schools, or vice versa. Then the analy­sis which keeps the groups separate w ­ ill show no effectiveness of school characteristics, ­because for each racial or ethnic group, the schools are uniformly bad or good.13 Other data in his report do suggest that black p­ eople go to very dif­fer­ent schools than white ­people. Only 16 ­percent of students are white in the average elementary school attended by black c­ hildren, whereas 91 ­percent of fellow students are white for the average white elementary student.14 Coleman nonetheless justifies his analy­sis by stating that failing to control for race would bias the analy­sis of schools ­because race has an effect on test scores in­de­pen­dent of any school effects. That conclusion, however, remains to be proven, and his methods do not allow him to prove it, ­because ­children are not randomly assigned to schools irrespective of their race.15 To see this from another perspective, consider an imaginary case. In a hy­po­thet­i­cal world, t­ here are high-­quality heart surgeons, with very high patient success rates, and other ­people who perform heart surgery but lack the proper training and skills, with very low patient success rates. The success rates of surgeons, however, are not publicly disclosed—­only their experience, mea­sured in years, which does not predict success. Patients nonetheless infer quality based on price and reputation, which is highly correlated with quality. The rich buy high-­quality surgery, whereas the poor pay as much as they can scrape together and hope for the best.

Unequal Access to Education  •  121

Now give ­these hy­po­thet­i­cal data to James Coleman and tell him to run a regression the way he did in his famous report. He would use patient mortality as the dependent variable and ­family income and experience of the surgeon as the in­de­pen­dent variables. The results would show that ­family income explains almost all of the variation in patient mortality ­a fter surgery. The media headline would be: “Go to Any Surgeon You Want, B ­ ecause Your Income Is All That ­Matters.” The “evidence-­based” policy that would come from this would be exactly wrong and 100 ­percent in­effec­tive: Give money to poor ­people but have them use the same surgeons they would have other­wise—­telling them that money, not surgeon skill, ­matters for survival. They would still die at the same rate as before. With education, we have done something similar. Since the Coleman report, the national teaching strategy has been based on the idea that poor kids ­won’t learn u­ ntil their f­ amily dynamics and socioeconomic status improve. The quality of teaching d­ oesn’t m ­ atter. That view is not universally held, of course, and many proponents of education reform (such as former Washington, DC, education chancellor Michelle Rhee) strongly disagree with that position, but it is arguably the dominant position of teachers’ ­unions, Demo­cratic Party leaders such as Hillary Clinton and Bernie Sanders, and far-­right white nationalists, who would other­wise have l­ ittle in common po­liti­cally. A second reason the conclusions of the Coleman report are wrong is that they are based on the wrong mea­sures of quality. Coleman’s mea­sures of teacher quality included teacher verbal test scores and other administrative data—­such as years of experience, race, and teacher education, none of which is consistently predictive of objective or subjective per­for­mance mea­sures, which ­were not part of the study. ­There is strong evidence that getting the mea­sure of quality right ­matters. Modern education research employs several sophisticated and accurate mea­ sures of teacher quality. One is to calculate the test score gains by the teacher’s average student. This is called the teacher’s value-­added to test scores. One large-­scale study finds that being taught by a high-­quality teacher for two years between ages 14 and 16 compared to a low-­quality teacher predicts a large increase in subject-­matter test scores, which are crucial to gaining ac­cep­tance for higher education levels.16 Moreover, this value-­added mea­sure is unrelated to the types of administrative mea­sures used in the Coleman report, such as teacher experience, education, and salary.

122 • Chapter 5

A third prob­lem is that the Coleman report did not track learning—or changes in test scores over time. Even using his own methods—­estimating the f­ actors that are correlated with an outcome—­would have shown two ­things: Student learning between 9th grade and 12th grade is substantial, and it varies systematically across schools. Th ­ ose are the results of a major study, called “Proj­ect Talent,” of 10,000 high school students that took place around the same time as the Coleman study.17 Fi­nally, and importantly, Coleman made no effort to estimate the causal effect of school, teacher, or neighborhood quality on achievement, but still inferred that ­these ­things do not ­matter much. To estimate the causal effect of teacher quality, you need two ­things: a valid mea­sure of teacher quality and a research strategy that uses some form of random assignment of students to teachers to account for the fact that f­ amily privilege or prior test scores are likely to be impor­tant predictors of access to high-quality t­ eachers and schools. Modern research has done this.

THE EFFECTS OF HIGH-­Q UALITY PUBLIC EDUCATION ON COGNITIVE ABILITY

Many studies in the field of psy­chol­ogy have examined how education affects IQ. It is obviously not enough to simply observe that ­people with higher education have higher average IQs (they do), b­ ecause natu­ral ability is likely to provide extra encouragement and motivation to pursue higher levels of education. Psychologists and other social scientists have tried to get around this prob­lem by observing how IQ changes over time for a given individual as he or she acquires more education and how IQ changes in response to government policy changes that encourage or mandate higher levels of education. A recent meta-­analysis by Stuart J. Ritchie and Elliot Tucker-­Drob examined 42 databases covering 600,000 p­ eople with IQ scores. They found that a year of education raises IQ between 1 and 5 points (which amounts to .07 to .33 standard deviations).18 The average effect across studies was 3.5 IQ points.19 To put that in perspective, the test score gap between non-­Hispanic white ­people and black p­ eople is approximately 15 points, according to Department of Education data from the National Assessment of Education Pro­gress; that amounts to four years of education, based on this analy­sis.

Unequal Access to Education  •  123

Results from the Proj­ect Talent database of 10,000 randomly selected U.S. high school students found similar effects of education on IQ mea­sured in the 1960s. For example, boys gained 14.4 IQ points in math and 19.1 points in lit­er­a­ture over three to four years of high school. Some IQ tests attempt to mea­sure cognitive ability that is less tied to knowledge and more about pure reasoning and memory. Even ­these tests reveal huge gains resulting from education. Among boys, memory increased 6.3 IQ points, abstract reasoning increased 7.8 IQ points, and visualization in three dimensions increased 9.5 points.20 To r­ eally appreciate the relationship between education and IQ, however, one needs to consider the massive gains in IQ over the last c­ entury. During that time, in the United States and ­on every continent on earth, ­people have gotten steadily better at IQ exams—at a rate of around 0.3 points per year. The phenomenon is known as the Flynn effect, ­after the New Zealand–based scholar who, in painstaking detail, documented a half-­century of gains in his groundbreaking 1984 article.21 Since then, the finding has been replicated around the world across ­every kind of intelligence test mea­sure.22 Indeed, for the Embu tribe in rural ­Kenya, IQ scores went up 11 IQ points from 1984 to 1998 for seven-­year-­old ­children of subsistence farmers on a mea­sure of intelligence (Raven’s matrices) that uses no words and is meant to mea­sure raw capacity for abstract reasoning (fluid intelligence).23 The K ­ enyan c­ hildren also showed large gains in verbal intelligence. The authors investigated potential ­causes and showed that the 1988 group had greater access to calories, particularly protein, but was more likely to suffer from intestinal parasites, even though less likely to suffer from hookworm. More conclusively, the 1988 group was more likely to have attended preschool (15 ­percent versus 7 ­percent), and Sunday school attendance also increased from 90 ­percent to 99 ­percent. ­Fathers ­were less likely to be living with the ­family in the latter group, and ­mothers less likely to be married, but maternal literacy and education ­were much higher, suggesting an enriched home environment. Altogether, the Flynn effect shows that the environment plays a massively impor­tant role in shaping intelligence, and IQ and cognitive exams should not be taken as mea­sures of natu­ral intellectual capacity. Rather, they mea­ sure intellectual per­for­mance that has been deeply ­shaped by environmental and social influences.

124 • Chapter 5

The other crucial insight of the Flynn effect is that intelligence is highly dynamic across cohorts. Population groups with relatively low IQ scores t­ oday would score the same as the average white American living in the 1960s and much higher than white p­ eople born at the beginning of the twentieth century. Looking back at the massive surge in invention and innovation in the nineteenth and early twentieth c­ enturies, it would be ridicu­lous to say that p­ eople ­were dramatically less capable than ­people living ­today. In the United States, the most inventive period in history began in roughly 1870 and lasted ­until roughly 1940, as mea­sured by the number of patents granted to U.S.-­based inventors per capita.24 It is bizarre to think that intelligence mea­sures would have suggested widespread stupidity. Th ­ ese p­ eople, most of whom had not attended college and would have done poorly on formal exams, nonetheless in­ven­ted the modern world. On many tasks not mea­sured on IQ tests—­such as mechanical and agricultural knowledge, skill at inventing and fixing tools and machines—­I suspect that p­ eople t­ oday would score lower. Quite likely, even the purest IQ tests mea­sure ele­ments of formal learning, as well as exposure to language and culture. P ­ eople adapt their learning and patterns of thought to the opportunities and demands of their environment. If parents and teachers emphasize the importance of lit­er­a­ture, writing, and abstract reasoning, ­children respond by developing t­ hose skills, and if po­liti­cal circumstances devalue ­those skills or cut off their applicability, investment in education w ­ ill be diminished. In any case, the link between IQ gains and mass schooling is clear. Enrollment, literacy, and educational attainment have risen dramatically since the early twentieth c­ entury (figure 5.1). The IQ data suggest that t­ hese investments in education have had enormous benefits to the population in terms of cognitive capacity and per­for­mance. Of course, other ­factors not strictly tied to education have also likely played a role: improvements in healthcare, sanitation, ­water quality, and exposure to toxins, for example. I also believe that the rise of more cognitively demanding occupations that require sophisticated literacy and numeracy on a regular basis—­especially in professional services—­have led to more concerted efforts for teachers, parents, and individuals to invest in the skills and way of life, such as daily reading and writing, consistent with ­those work opportunities.

Unequal Access to Education  •  125 120 100

100 80

78.92 69.31

87

84.4

74.2

60

96.5

88.22

40 20 0

1909 IQ

1940

1970

2008

Share of population enrolled in school, 5–17 year olds

Figure 5.1. ​A ­century of IQ and education gains in the United States, 1909–2008 Sources: I simulated IQ scores using data from Jakob Pietschnig and Martin Voracek, “One Century of Global IQ Gains: A Formal Meta-Analysis of the Flynn Effect (1909–2013),” Perspectives on Psychological Science 10.3 (2015): 282–306, which show average annual gains of roughly 0.31 IQ points from 1909 to 2008 in the United States. This is consistent with research from James R. Flynn and Lawrence G. Weiss, “American IQ Gains from 1932 to 2002: The WISC Subtests and Educational Progress,” International Journal of Testing 7.2 (2007): 209–24. Education data is from Table 8 in Thomas D. Snyder, “120 Years of American Education: A Statistical Portrait,” https://nces​ .ed.gov/pubs93/93442.pdf. I calculated the 2008 data for educational attainment using enrollment data from the 2006–2010 Five-Year American Community Survey, using American Fact Finder and taking the mean enrollment values across ages 5–17.

THE ­C AUSES OF NEGATIVE-­F LYNN EFFECTS

Recently, scholars of intelligence have pointed to some evidence for a reversal of the Flynn effect—­a negative-­Flynn effect, meaning a decline in intelligence between cohorts at the population scale. Thus far the evidence is thin, short in duration, and applicable to only a small number of countries.25 The most recent large-­scale lit­er­a­ture review finds a continuation of IQ gains in e­ very continent in recent de­cades, but t­ hese studies cobble together many very small samples, often from nonrepresentative populations.26 Data from large-­scale international cognitive assessments provide the relevant evidence ­either way, with the most noteworthy being the OECD’s Programme for International Student Assessment (PISA) for 15-­year-­olds. The PISA exam has been conducted from 2000 to 2015 with comparable scales for ­people of the same age in a large number of countries. The data show no significant change in math, science, or literacy scores over vari­ous years

126 • Chapter 5

during this period for the average OECD student. In OECD publications, reported math trends cover the longest period (2003–2015), and only a handful of the 63 countries saw a significant decrease: Finland, Australia, Iceland, Netherlands, Belgium, Sweden, Canada, France, Hungary, the Slovak Republic, the Czech Republic, and Vietnam.27 Many countries saw significant math test s­ core gains over this period, including Italy, Brazil, Indonesia, Israel, Portugal, Peru, Costa Rica, Argentina, Rus­sia, Poland, Romania, and Albania. The upshot is that for most rich countries, the Flynn effect appears to have leveled off over the last 15 years, but t­ here is no evidence for decline outside of a small number of countries. Norway and Finland are two countries where ­there is convincing evidence of a decline in cognitive per­for­mance. Large-­scale data from Finland shows test scores having fallen, which is consistent with Finland’s PISA data. The test scores of Finnish military conscripts grew over the late 1980s and early 1990s and peaked in 1997 (representing people born in 1977 or 1978); by 2009, scores were approximately 2 IQ points lower.28 A strikingly similar pattern emerged in Norway, where the test scores of Norwegian-­born military conscripts r­ ose by 3 IQ points for t­ hose born between 1962 and 1975 and then fell by roughly 2.5 IQ points for ­those born between 1989 and 1991.29 Thereafter, IQ scores have apparently stabilized in Norway or even ticked back up. C ­ hildren born in 2000 scored slightly better than ­those born in 1988—­with differences falling within the margin of error—on the PISA math exam, and reading scores increased significantly for c­ hildren born in 2000 compared to ­those born in 1994. Some hereditarian IQ scholars attribute the decline to biological dysgenic effects (higher fertility rates among lower-IQ parents), but the data from PISA suggest that that is highly unlikely, and an analy­sis of the Norwegian data rules it out entirely. More-­ educated parents—­ with higher IQs, typically—­have been having fewer c­ hildren than less-­educated parents in essentially ­every country, and yet the positive Flynn effect has dominated in all but a few, and the only declines have occurred for ­those born since 1980.30 If dysgenic effects m ­ atter at all, they must be trivial compared to environmental forces such as access to education or better health. Moreover, I find that the negative trend in Finland and the positive trend in Norway for PISA results occur for both the native-­and foreign-­born student populations in ­those countries, which, again, makes dysgenic ­factors unlikely, since the fertility patterns of immigrants differ from ­those of native northern Eu­ro­pe­ans.31

Unequal Access to Education  •  127

The c­ auses of the positive and negative Flynn effects w ­ ere teased out more systematically by Bernt Bratsberg and Ole Rogeberg. They isolated test results from Norwegian-­born males who had b­ rothers who also took the test (355,438 individuals) and found that both the positive and the negative Flynn effect occurred in the same families, depending on when the ­brothers w ­ ere born, and matched the Flynn effects that occurred across dif­fer­ent families. The unavoidable implication is that the Flynn effect is caused neither by parenting characteristics that apply equally to ­brothers in the same ­family nor by genes, ­because each ­brother is equally likely to have more or less advantageous genes for intelligence. Instead, the Flynn effect observed in recent de­cades in Norway—­and the more recent smaller decline—is explained by cohort-­level environmental ­factors that work across families, such as the quality of neighborhoods or schooling.

THE EFFECTS OF QUALITY OF SCHOOLING ON IQ

The gains in IQ from education apply to the average student, with averagequality schooling. Since schooling ­matters, it stands to reason that higher-­ quality education—­either formally in school or through after-­school programs, tutoring, or parental efforts at home—­would result in even larger IQ benefits. Indeed, ­there is now irrefutable evidence showing that higher-­quality education results in greater cognitive gains. One of the most impressive studies collected data on 2.5 million U.S. students from 1989 to 2009 and linked ­these data to teacher and administrative rec­ords in one the country’s largest school districts. The researchers—­ Raj Chetty, John Freidman, and Jonah Rockoff—­calculated a valid mea­sure of teacher quality—­effectiveness at raising test scores—­and used quasi-­ experimental methods to link teachers to students by taking advantage of random changes in teacher staffing.32 They find that a one standard deviation in teacher effectiveness raises student test scores by .14 and .10 standard deviations for students in grades three through eight (corresponding, roughly, to ages 9 to 14). In a follow-up paper that links students to income tax rec­ ords, the same authors find that even one year of having a teacher that is one standard deviation above the mean raises lifetime earnings by approximately $39,000.33 One can imagine that the cumulative effects of having consistently good or bad teachers would be dramatic.

128 • Chapter 5

Another approach to mea­sur­ing teacher quality does not rely on test score data to mea­sure quality. Instead, it relies on classroom observations from in­de­pen­dent observers who are trained to use certain criteria. ­There are several dif­fer­ent mea­sures that have been studied in detail, and teachers tend to be ranked in very similar ways using any of the mea­sures. Yet, ­these mea­ sures clearly capture something distinct from the test ­score gains of previous students, b­ ecause they are only moderately correlated with them. One such observational mea­sure of teachers—­called CLASS—­seeks to mea­sure the quality of student-­teacher interactions. It mea­sures emotional support, the quality of instruction, and the quality of classroom organ­ization and management.34 A teacher’s CLASS rating strongly predicts gains in student math and language, as well as noncognitive skills related to student effort. When combined with student survey information about their teacher, ­these ratings are especially effective, yielding large increases in both learning and student effort between good and not-­so-­good teachers.35 We know ­these effects are causal ­because students ­were randomly assigned to classrooms in this study. Moreover, the findings have been replicated by other scholars in a dif­f er­ent sample of 24,000 kindergarteners. They found that the result of assigning students to better teachers as rated by CLASS was significant improvement in math and reading, as well in as the c­ hildren’s executive functioning—­ including their ability to pay attention and regulate their emotions.36 Other data using CLASS at the prekindergarten level is publicly available for scholars from the Study of Statewide Early Education Programs, and I have analyzed it.37 Using informal mea­sures of knowledge, such as letter, number, and color recognition, as well as counting, I found that the black-­white gap in test scores among pre-­K to fourth-­grade students is roughly nil. Black parents seem to do just as well as more-advantaged and higher-educated white parents at teaching their ­children ­these t­ hings (­table 5.1). Asian ­children came in the most prepared, and Hispanic c­ hildren the least prepared. On more formal IQ tests of math, vocabulary, and language, the results ­were rather dif­fer­ent. Asian, Black, and Hispanic ­children scored significantly lower than white p­ eople and multiracial ­children. The mean IQ gap was 0.91 between white p­ eople and black p­ eople and 0.82 between white p­ eople and Asians. ­There ­were no significant IQ differences between Asians and black ­people. As expected, differences in parental education and income explain much of t­ hese gaps.

Unequal Access to Education  •  129

­ able 5.1.  Fall pre-­K indergarten 4 test scores by race and ethnicity, T expressed as standardized scores with mean of zero and standard deviation of one

White Multiracial Asian Native American African American Latino/a

Formal IQ tests

Knowledge tests

0.42 0.10 −0.39 −0.47 −0.48 −0.58

0.28 0.14 0.95 −0.37 0.17 −0.50

Source: Data from Study of State-­wide Early Education Programs (SWEEP) from The National Center for Early Development and Learning (NCEDL) at UNC Chapel Hill. Test scores w ­ ere collected in fall and spring of pre-K–4. Sample size is 721 classrooms with 2,982 c­ hildren across 11 states. Summary means presented h ­ ere use fall test scores and fall sample weight. Formal IQ mea­sures include the Peabody Picture Vocabulary, Woodcock to Johnson III, and the Oral and Written Language Scale. Knowledge tests use number counting, number recognition, letter recognition, and color recognition. For details see Diane Early et al. “Pre-­K indergarten in Eleven States: NCEDL’s Multi-­State Study of Pre-­K indergarten and Study of State-­Wide Early Education Programs (SWEEP).” ICPSR34877-­v1. Ann Arbor, MI: Inter-­university Consortium for Po­liti­cal and Social Research [distributor], 2013-10-02. http://­doi​.­org​/­10​.­3886​/­ICPSR34877​.­v1.

THE EFFECTS OF QUALITY OF TEACHING ON EARLY CHILDHOOD IQ GAPS

In examining CLASS data, I also found that students learn more when they are assigned to higher-­quality classrooms, and this has impor­tant implications for cognitive gaps by race and class. Black and Hispanic pre-­K–4 students demonstrated extraordinary gains in learning and IQ when placed in high-­ quality classrooms. Black ­children gained roughly one-­third of a standard deviation in language and math over just a few months of instruction when they w ­ ere placed in high-­quality classrooms, but they experienced almost no relative gains in low-­quality classes (­table 5.2). Likewise, Hispanic ­children demonstrated very large learning gains in high-­but only modest gains in low-­ quality classrooms. For the average white student, classroom quality did not seem to make much of a difference in how they scored relative to their peers. The reason may be that the white parents of ­children in lower-­quality classrooms w ­ ere more apt to engage in supplemental learning exercises at home, but the data do not provide any clues.

130 • Chapter 5

­ able 5.2.  Learning gains from fall to spring in high-­versus low-­ T quality classrooms among pre-­k indergarten–4 students, mea­sured in standard deviation units High-­quality classrooms

Low-­quality classrooms

0.37 0.34 0.21

0.06 0.04 −0.04

0.39 0.13 0.12

0.16 0.17 0.03

0.15 0.00 0.02

0.15 0.08 −0.01

Black students Change in language IQ Change in math IQ Change in overall cognitive scores

Hispanic students Change in language IQ Change in math IQ Change in overall cognitive scores

White students Change in language IQ Change in math IQ Change in overall cognitive scores

Source: My analy­sis of data from Study of State-­wide Early Education Programs (SWEEP) from The National Center for Early Development and Learning (NCEDL) at UNC Chapel Hill. Asians ­were only 2.7% of the sample and so are omitted ­here for lack of reliable estimates. High-­quality classrooms are in the top quartile, while low-­quality are in the bottom quartile for the summary CLAS mea­sure. Test scores ­were collected in fall and spring of pre-K–4. Sample size is 721 classrooms with 2,982 c­ hildren across 11 states. Summary means presented h ­ ere use Spring sample weight. Language IQ mea­sures by the Peabody Picture Vocabulary; math uses Woodcock to Johnson III, and overall score combines the two above with the Oral and Written Language Scale and assessments of number counting, number recognition, letter recognition, and color recognition. For details see Diane Early et al. “Pre-­K indergarten in Eleven States: NCEDL’s Multi-­State Study of Pre-­ Kindergarten and Study of State-­Wide Early Education Programs (SWEEP).” ICPSR34877-­v1. Ann Arbor, MI: Inter-­university Consortium for Po­liti­cal and Social Research [distributor], 2013-10-02. http://­doi​.­org​/­10​.­3886​/­ICPSR34877​.­v1.

While t­ hese results are adjusted for the relative per­for­mance of same-­age peers, students from all racial backgrounds progressed on raw mea­sures of learning over the course of the school year. By spring, the average pre-­K–4 student could recognize and name more letters and numbers, count higher, and identify more colors. All of ­these ­things ­were positively correlated with higher IQ, which also increased. Raw IQ scores went up by roughly .41 to .49 standard deviations (6.2 to 7.5 IQ points) relative to fall scores.38 Thus, even average-quality classrooms raise IQ substantially for all races, but the best

Unequal Access to Education  •  131 0.3

Standard deviation units

0.2 0.1 0 –0.02

0.20

–0.1 –0.16

–0.2 –0.3 –0.4

–0.35 African American

Hispanic

Multiracial

White

Figure 5.2. ​Pre-K–4 classroom quality by race and ethnicity Source: My analy­sis of SWEEP database, using fall student weight. See note in table 5.1 for details.

teachers are especially effective in raising scores for c­ hildren from less-­ privileged backgrounds and thereby lowering the overall achievement gap between groups. This is exactly what you would expect if high-­quality teaching compensates for ­limited ­family exposure to reading and literacy promotion for low-­income c­ hildren.39 High-­quality teaching allows disadvantaged students to catch up with their more advantaged peers. Th ­ ese teacher quality effects go a long way to explaining longer-­term gaps in cognitive per­for­mance in late adolescence and adulthood. Black c­ hildren, especially, but also Hispanic c­ hildren w ­ ere much more likely than white c­ hildren to be in low-­quality classrooms (figure 5.2). The black-­white pre-­K gap in classroom quality is just over half a standard deviation, and that translates into missed learning opportunities for black p­ eople and divergence rather than convergence in IQ. Other research from Washington State also finds that black ­people, Hispanics, and low-­income c­ hildren are consistently assigned to lower-­quality teachers, as mea­sured in several ways.40 Fortunately, ­there is strong evidence that teachers can be taught to manage their classrooms more effectively and literally become better teachers.41

132 • Chapter 5 THE POSITIVE EFFECTS OF SCHOOL-­B ASED INTERVENTIONS ON BOOSTING IQ

Above, I mentioned just a few examples of recent studies using rigorous experimental designs that have proven that high-­quality schooling raises cognitive ability and the personality traits associated with higher income and health, but t­ hese examples are hardly exhaustive. One of the world’s top education scholars, economist Roland Fryer, compiled evidence from 196 education experiments, in which c­ hildren ­were randomly assigned to dif­fer­ent treatment conditions, ranging from early childhood education, family-­based interventions, and school-­quality interventions. The home-­based interventions in Fryer’s study had small effects on reading and math scores, with some impor­tant exceptions. Reducing poverty and giving parents extra educational resources w ­ ere rather in­ effec­tive. On the other hand, home visits to low-­income ­mothers from a specialized nurse during and ­after pregnancy proved very effective at boosting maternal and child health outcomes, as well as cognitive ability. A random experiment showed that 12-­year-­olds who participated in the program as infants scored 0.20 standard deviations higher on IQ tests than their peers in the control group.42 Aside from early childhood interventions, getting parents involved more in their child’s school has also proven to be effective. Overall, however, school-­based interventions have been more consistently power­ful. Among the most effective education treatments are intensive (“high-­dosage”) tutoring, which has huge effects on test scores; attendance at high-­quality charter schools; and the professional development of teachers (­table 5.3). Attendance at high-­quality pre-­K classrooms has also proven to be highly effective at raising IQ according to the most rigorous social science research.43 High-­quality childcare at an early age emphasizes language learning through games and has been found to eliminate IQ gaps between very disadvantaged ­children and their average American peers. The Infant Health and Development Program had a similar effect on low-­birth-­weight, mostly minority ­children, and most of the IQ gains persisted for years a­ fter the program ended.44 Similar positive long-­term effects from the Chicago Child-­Parent Center ­were found to persist in terms of education and the probability of arrest for the ­children who attended it, who w ­ ere mostly black.45

Unequal Access to Education  •  133

­Table 5.3.  The experimental effects of educational interventions

Early childhood interventions All home-­based interventions Parental involvement Educational resources Poverty reduction All school-­based interventions Student incentives Tutoring (high-­dosage) Tutoring (low-­dosage) Teacher certification Teacher incentives General teacher professional development Managed teacher professional development Data-­driven teacher evaluations Extended time School choice/vouchers Charters “No Excuse” charters

Math

Reading

0.12 0.04 0.12 −0.06 0.01 0.14 0.04 0.4 0.07 0.03 0.05 0.17 0.06 0.11 −0.03 0.08 0.12 0.17

0.2 0.08 0.14 0.07 0.02 0.2 0.1 0.41 0.05 0 0 0.15 0.49 0.07 0.16 0.07 0.07 0.1

Note: Average treatment effect is reported above in standard deviation units for 196 studies that use random assignment or experimental methods. High-­dosage tutoring means small groups (six or fewer) for four days or more per week. Source: Reproduced from Fryer, Roland G. “The Production of ­Human Capital in Developed Countries: Evidence From 196 Randomized Field Experiments.” Handbook of Economic Field Experiments 2 (2017): 95–322.

In consistently demonstrating large effects on educational achievement, t­ hese interventions prove that public education has enormous potential to enhance the cognitive ability of ­children born in low-­income h ­ ouse­holds, especially when combined with f­ amily interventions that promote the adoption of routine healthy be­hav­iors.

THE EFFECTS ON SKILL OF GROWING UP IN A RICH REPUBLIC: EVIDENCE FROM IMMIGRATION

Across Eu­rope and the United States, far-­right politicians and supporters view immigration as a threat to the prosperity of the country. They seem to believe that immigrants—­who typically come from poorer countries with lower rates of educational attainment—­are not capable of making productive

134 • Chapter 5

contributions to the host country. In the most extreme cases, opponents of immigration argue that the cultural or ge­ne­tic characteristics of immigrants make them and their c­ hildren incapable of contributing. Th ­ ese commentators, encouraged by recent comments by President Trump, appear to regard the poverty of a country as an indicator of the natu­ral capacity for its ­children to learn and develop.46 They are wrong. Something almost incredible happens when ­children from poor countries with very low-­quality education systems grow up in countries with strong and equitable educational systems.47 A comprehensive 2002 study of per­for­mance in En­glish national exams taken by half a million ­children found that African immigrants to E ­ ngland born in 1991 scored only 2 IQ points ­behind white ­people at ages 7 and 11, and Indian c­ hildren scored the same as white En­glish children at age 7 and somewhat higher at age 11.48 Both groups significantly outscored Pakistani and Bangladeshi immigrants, but they scored lower than Chinese immigrants. Black African immigrants scored higher than black Ca­rib­bean immigrants. But t­ here is no natu­ral law that group test scores are stable. For example, exam data from the London borough of Lambeth (a poor area of London with a high proportion of immigrant families) for 2011 show that the highest-scoring ethnic groups w ­ ere black African speakers of Luganda (from Uganda) and Krio (from Sierra Leone), who had higher pass rates on the national exams than Chinese immigrants.49 Nigerian Igbo and Yoruba speakers, as well as Ghanaians, outperformed Indian immigrants and white British natives (­table 5.4). The high a­ chievement of t­ hese African ethnic groups—­who come from the nations that supplied much of the Atlantic slave trade—­cannot be attributed to high socioeconomic status. Many of the students are eligible for ­free school lunches ­because their families have low incomes, but, like East Asians and Indians, their parents are highly motivated for them to study and succeed in education, and they behave accordingly. And in Lambeth, at least, the qualitative evidence suggests that the teachers and staff provide a high-­quality learning environment.50 While not all ethnic groups from Africa are so high achieving, by 2013, the average 16-­year-­old black African in E ­ ngland was scoring higher than the average white British student on the national exams.51 Likewise, ­children with one black African parent and one white parent w ­ ere also

Unequal Access to Education  •  135

­Table 5.4.  Percentage of 16-­year-­olds with passing grade (5+A*-­C) or higher on a­ ll five of England’s Key Stage 4 national exams, 2011 Black African, Luganda s­ peakers Black African, Krio speakers Chinese Black African, Igbo speakers Black African, Yoruba speakers Black African, Ga speakers Indian White British Bangladeshi Black African Pakistani Black Ca­rib­bean

83% 78% 76% 76% 75% 75% 72% 55% 54% 53% 50% 44%

Source: Feyisa Demie, “Raising the Achievement of Black African Pupils: Good Practice in Schools” (Lambeth Research and Statistics Unit, 2013), available at https://www.lambeth​ .gov.uk/rsu/sites/lambeth.gov.uk.rsu/files/Raising_the _Achievement_of_Black_African_Pupils-Good_Practice _in_Schools_2013.pdf, accessed June 18, 2019.

outscoring white p­ eople, achieving the same scores as t­ hose with two African parents. White British and black Africans both outscored black Ca­rib­be­ans.

THE EFFECTS OF MOVING TO A BETTER COUNTRY

To further understand how immigration might affect IQ, I examined data from the Programme for the International Study of Adult Competencies (PIAAC). With this initiative, the OECD has created the most ambitious international database ever assembled for the study of adult cognitive per­for­ mance across countries, examining a random sample of 250,000 ­people aged 16 to 65 living in 33 mostly rich countries. The IQ of immigrants who arrive in developed countries before the age of ten is roughly 97, just one-­fi fth of a standard deviation below the international norm of 100. Scores are much lower—by as much as 10 IQ points—­

136 • Chapter 5 97.0

96.7

90.7

90.5

90.2

87.2

1 to 4

5 to 9

87.5

87.6

10 to 14 15 to 19 20 to 24 25 to 34 35 to 44 45 to 65

Figure 5.3. ​Mean IQ of immigrants to rich countries by age of arrival when tested as adults Source: Author analysis of OECD PIAAC data for 10,808 adults in 22 countries, which is everyone in the database with nonmissing data for age of entry and immigrant status. IQ scores were first adjusted by age using AGEG10LFS, which has five categories, before putting into standardized IQ units of mean 100 and standard deviation of 15.

for ­t hose who enter the country as adults (figure  5.3). If se­lection bias explained the high per­for­mance of immigrants (meaning that only highly capable parents migrate), then it should not ­matter at what age their ­children arrive, but if the environment ­matters, then the amount of time spent in the poor country of origin compared to the rich destination country should ­matter, and it does. Other scholars have found the same effect, even for c­ hildren within the same ­family. For immigrants to Germany, arriving at age one compared with ­doing so at age 15 raises educational attainment by two years. Similar results have been found for immigrants to Sweden and Norway.52 Arriving before age ten seems to be particularly beneficial.53 I believe the reasons for the age pattern are twofold: First, rigorous classroom evaluations show that school quality is dramatically better in rich countries, so arriving in a rich country early means more cumulative years of higher-­quality education; second, early childhood education is especially power­f ul relative to adult education for practical reasons—­children a­ ren’t expected to work and take care of families—­ and, likely, for biological reasons—­brain development is more rapid among ­children and levels off during puberty.54

Unequal Access to Education  •  137

Table 5.5.  Education, entrepreneurship, and income of African immigrants growing up in the United States by age of arrival

Moved to United States at 10 or older Moved to United States before age 10 Whites, born in United States (aged 30–50)

Percentage with at least a bachelor’s degree

Percentage with a doctorate or professional degree

Percentage who own an incorporated business

Median income

41.6

5.1

2.9

$26,134

53.5

8.8

4.3

$39,000

36.3

3.4

3.1

$36,588

Source: Data from the 2012–2016 American Community Survey. Analy­sis is ­limited to adults aged 30 to 50 between 2012 and 2016. ­There w ­ ere 31,610 African immigrants surveyed, 1,532 of whom arrived before age 10.

Test scores from 15-­year-­olds who took the OECD’s PISA exam show the same effect. Only a handful of highly developed countries reveal the country of birth of test-­takers. Among them are New Zealand, Finland, Denmark, and Belgium. ­Children born in Africa who arrive in ­these countries ­after age ten score well below international norms (91.2), but t­ hose who arrive before age ten score above international norms (101.1). For a black African, being raised in northern Eu­rope or New Zealand instead of Africa is akin to gaining 33 years from the Flynn effect. This applies just as well to other mea­sures of success. African immigrants who come to the United States from poor countries before the age of ten are far more likely to have a college degree, own a business, and earn higher incomes as middle-­aged adults—­aged 30 to 50—­than ­those who arrive ­after the age of ten. In fact, the ­children of African immigrants earn higher incomes than U.S.-­born white p­ eople and are more than twice as likely to earn a PhD or advanced professional degree (­table 5.5). Indeed, c­ hildren born in poor countries with low national mea­sures of educational attainment or cognitive per­for­mance routinely become more educated than U.S.-born white ­people and Asians. One out of ­every four ­children (25 ­percent) of Ira­nian immigrants who arrived in the United States before age ten had obtained a PhD or advanced professional degree between the

138 • Chapter 5

ages of 30 and 50, making American-­raised Ira­nian immigrants the highesteducated group of Americans. Immigrants from 16 other countries ­were more likely to obtain this level of education than native-­born Asian Americans (10.3 ­percent), and immigrants from 51 countries surpassed native-­born white ­people (3.4 ­percent). Th ­ ese high-­performing countries cover much of the world’s regions: India, Taiwan, Israel-­Palestine (which are grouped together in the data), South Africa, Nigeria, Pakistan, the United Kingdom, Malaysia, Syria, Egypt, Australia, ­Korea, Lebanon, and Ireland. Even more countries would be represented, but I l­imited the analy­sis to countries with at least 100 respondents surveyed by the census to minimize sampling errors. Taken together, the evidence is conclusive that educational opportunities in rich countries are far better than t­ hose in the home regions most mi­g rants are leaving, and this exposure to high-­quality education leads to massive gains in the skills and knowledge they acquire and their ability to be productive citizens.

REASSESSING THE HERITABILITY OF INTELLIGENCE: FROM TWIN STUDIES TO GENOMIC STUDIES

In 1994, a group of 52 scientists working in psy­chol­ogy and other areas on the mea­sure­ment and inheritance of intelligence signed a letter in the Wall Street Journal stating that “individuals differ in intelligence due to differences in both their environments and ge­ne­tic heritage. Heritability estimates range from 0.4 to 0.8 (on a scale from 0 to l), most thereby indicating that ge­ne­tics plays a bigger role than does environment in creating IQ differences among individuals.”55 Moreover, leading ge­ne­tics scholar Robert Plomin has recently suggested on the basis of twin studies that the heritability of adult IQ is as high as 80 ­percent and that both schools and families make l­ ittle difference in accounting for IQ.56 Yet ­there are good reasons to believe that ­these conclusions are badly overstated. Very high heritability does not imply that environment has no impact. Nor does it indicate w ­ hether genes or environment has the strongest effects.57 In fact the latest evidence suggests that it is the environment that is the dominant ­factor in accounting for differences between individuals, and as I ­will

Unequal Access to Education  •  139

argue in the next chapter, the environment is almost certainly the only significant source of differences between groups of ­people. So why has heritability been considered so impor­tant? The answer lies, in large part, in the dependence upon twin studies. Typically, twin studies estimate heritability of IQ and other traits by observing the difference in correlations between identical twins and nonidentical same-­ sex sibling twins. (It is also worth noting that using opposite-­sex nonidentical twins biases the heritability estimate since gender has significant environmental consequences.) Early estimates, which ­were typically based on small samples, often produced highly variable results. The most compelling estimates are from meta-­analysis. Recently, a team of ge­ne­tics scholars led by Tinca Polderman analyzed the results of nearly e­ very twin study ever published in a massive meta-­analysis of hundreds of thousands of twin pairs across many countries. For male identical twins, they found that cognitive ability had a moderately high correlation of 0.56; this would seem to imply that 31 ­percent (or 0.56 squared) of the variation in cognitive ability is explained by having equivalent genes. For nonidentical male twins, the correlation was 0.34. Th ­ ere are a few methods to calculate heritability based on ­these data. Using Falconer’s formula, scientists multiply the difference between ­these two correlations by two. The reason is that identical twins share all of their genes, whereas nonidentical twins (siblings) share only 50 ­percent of their genes, on average. Twice the difference yields a heritability estimate of 0.45 for IQ, which is somewhat higher than what the raw correlation between identical twins would suggest.58 Heritability estimates for cognitive ability fall somewhere in between ­those for health and height, when using the same database and methods used by Tinca Polderman and co-­authors. Male identical-­twin mortality is correlated at just 0.36, and male dizygotic twin mortality is correlated at 0.21. The Falconer’s formula suggests that mortality is 29  ­percent heritable.59 On the other end, male identical twins have very high correlations on height (0.92), compared to just 0.53 for dizygotic male twins. That suggests heritability of 78 ­percent for height. The estimates based on the meta-­analysis above use data from ­every country with published results, but the heritability estimates vary widely across countries. For male identical versus nonidentical twins, heritability is as low as 0.14 in Japan and 0.31 in Denmark. In the United States, estimates ­were 0.43; in Sweden, they w ­ ere 0.56. Thus, one cannot say t­ here is

140 • Chapter 5

one “true” figure for the heritability of intelligence; it depends very much on the date of the study, the sample population, and most of all, on the environment. This is also true for the heritability of educational attainment, according to another meta-­analysis from Amelia Branigan and ­others.60 Their estimates for educational heritability range from −5 ­percent to 77 ­percent. The g­ rand mean across all studies they located is 40 ­percent. This raises serious concerns about the validity of twin methods. In contrast to older studies with smaller sample sizes, their study found large effects for both shared and nonshared environment. While it is on the low range of the old consensus, the 45 ­percent estimate of IQ heritability from twin studies is certainly too high. Beyond IQ, it is well known among ge­ne­tics researchers that twin-­based methods lead to higher estimates of heritability than methods that analyze a­ ctual genes.61 ­There are two main reasons why twin studies generate higher heritability estimates. One reason f­avors twin studies. Insofar as the assumptions are correct, twin estimates provide a complete analy­sis of all ge­ne­tic effects—­ including gene-­to-­gene interactions and gene-­environment interactions. Indeed, t­ here has been considerable research on gene-­environmental interactions as summarized in a recent review by Richard Nisbett and co-­authors.62 Eric Turkheimer and his collaborators have found evidence that ge­ne­tic heritability—­based on twin studies—­tends to be lower in families with lower socioeconomic status and higher in more affluent families. This makes sense if one assumes that environmental variation is lower in affluent families.63 By contrast, studies of a­ ctual genes cannot take into account all of t­hese complexities using existing methods. Since h ­ umans have over 99 ­percent of their genomes in common, ge­ne­tic researchers focus on the locations where genes vary between individuals; they are called single nucleotide polymorphisms, or SNPs. Estimates of heritability from SNPs generally assume a ­simple relationship between a gene and a phenotype (e.g., cognitive per­for­ mance), the idea being that having a nucleobase of adenine instead of guanine, for example, at a par­tic­u­lar location is always better for intelligence, regardless of the environment or other genes pre­sent. In real­ity, ­whether it is good or not may depend on the presence of other complementary genes or environmental circumstances. Twin studies do not make this kind of assumption, but neither can they shed light on what ge­ne­tic mechanisms are working.64 The other reason twin studies show higher heritability estimates than gene studies is that they impose strong assumptions that bias the results upward.

Unequal Access to Education  •  141

The heritability formula assumes that nonidentical twins share the same environment and to the same extent as identical twins. That simply ­isn’t true. The fact is that identical twins share certain environmental experiences to a greater extent than nonidentical twins. As a recent review article concludes: “Identical twin pairs w ­ ere more similar than fraternal pairs on bullying, sexual abuse, physical maltreatment, emotional abuse and neglect, and general negative life events or trauma.” 65 The Polderman database likewise shows that negative life events in childhood are much more likely to be shared by identical than by dizygotic twins (the correlations are 0.55 versus 0.37).66 Identical twins are also more likely to be assigned to the same classroom than nonidentical twins and to spend more time with each other.67 In short, impor­tant life experiences are more closely linked among identical twins. Alternative twin-­based heritability methods, such as studying twins who ­were a­ dopted, also have impor­tant flaws. The convention is to assume that any association between the IQ of identical twins reared apart must be entirely ge­ne­tic, but that’s not right. The homes provided by a­ doptive parents are often very similar, especially in terms of parental education.68 Moreover, the womb is an environment with varying exposures to nutrition, toxins, and stress, and ­adopted ­children often spend significant time with their biological m ­ other before adoption—­sometimes years. The upshot is that even twins who are reared apart may share many impor­tant environmental circumstances in the womb and the first few months or years of life. Another problematic assumption in twin studies is that siblings share 50 ­percent of their ge­ne­tic material: This is correct only on average. In fact, siblings share between 37 ­percent and 62 ­percent of their genes, according to a study that examined the a­ ctual genomes of siblings.69 This variation should be taken into account when calculating heritability from twin studies. D ­ oing so, as scientists did recently, lowers the heritability estimates by about half. In the case of educational attainment, the heritability estimate falls from 43 ­percent to 17 ­percent.70 Given t­ hese limitations, the direct study of a­ ctual genes offers a compelling way to mea­sure heritability, despite the difficulties mentioned above. This method was impossible ­until recent breakthroughs in the mapping of the ­human genome made the study of molecular genomics feasible on a large scale. The three most ambitious scientific efforts to date have brought together teams of scientists from around the world. The first was published in 2013 by Cornelius Rietveld and a massive team of scholars at 129 dif­f er­ent research

142 • Chapter 5

institutions, who analyzed genomic data from 101,069 p­ eople.71 They looked at the correlation between over one million genes and educational attainment, ­after controlling for gender and year of birth, among a sample of p­ eople with Eu­ro­pean ancestry. Using this method, Rietveld and his co-­authors identified just three genes (SNPs) as having a statistically significant effect on education, and the strongest effect could explain only 0.02 ­percent of the variation in educational attainment, which is equal to about a month of schooling. When they added up each of the million tiny effects from the individual genes—to get a “polygenic score”—­they w ­ ere able to explain roughly 2 ­percent to 3 ­percent of the variation in educational attainment and IQ. They also suggested that with an extremely large sample size, the effect of each gene would become more accurate, and they would be able to explain 22 ­percent of the variation in educational attainment with fully mapped genomes from millions of p­ eople.72 The second study had access to a larger sample of p­ eople (293,723 instead of 101,069), and consistent with their prediction, the explanatory power increased. That effort, led by Aysu Okbay, Jonathan Beauchamp, and scholars from 185 research institutions, identified 74 genes that had significant associations with educational attainment.73 Yet, even a­ fter adding up the effects of e­ very gene (including t­ hose that a­ ren’t significant), their polygenic score could explain only 3.2 ­percent of the variation in educational attainment.74 Fi­nally, in a third, still larger analy­sis, the same team of scholars—­this time with James J. Lee, Robbee Wedow, and Aysu Okbay listed as the first of dozens of authors—­obtained genomes for 1.1 million individuals. This much larger sample did give them the power to identify more genes that significantly predict educational attainment (1,271 compared to 74 in the previous study).75 It is worth noting that many of t­ hese ge­ne­tic effects appear to be biased by ­family environment, since the predicted effect of 35 ­percent of genes points in the opposite direction when siblings are compared to one another. In other words, say that gene rs9859556 on the third chromosome predicts more education across 1.1 million p­ eople of Eu­ro­pean ancestry, but the results change when only siblings are compared. If two ­brothers are compared, the ­brother without the gene systematically has more education. This kind of switch in the predicted direction occurred for 35 ­percent of genes that w ­ ere identified as significant. This type of reversal of a regression coefficient occurs commonly

Unequal Access to Education  •  143

in social science when an omitted variable (such as ­family background) is added to a model that is fundamentally biased without it. This finding, in and of itself, directly contradicts Plomin’s contention that nonge­ne­tic ­family effects do not m ­ atter to educational outcomes. Additionally, the authors find impor­tant environmental effects on how genes are expressed. The genes identified as significant had dif­fer­ent effects in dif­fer­ent cohorts, which suggested to the authors that environmental circumstances—­such as access to public education and college or opportunities to work in cognitively demanding professional jobs—­interact significantly with genes. Nevertheless, in this larger sample, the polygenic score (the combined effect of all the genes) was again used to predict educational attainment, controlling for sex, age, and multicollinearity with closely related genes. The amount of variation explained by the polygenic score (the model’s R 2) was approximately 13 ­percent in the strongest model, which used complex statistical techniques to calculate both the polygenic score and its predictive power. The strength is somewhat lower if only genes with a significant individual relationship to educational attainment are included, and it falls to 4.6 percent if the authors control for parental education, which has a stronger effect than genes—­presumably ­because it includes both ge­ne­tic and nurture effects. Fi­nally, g­ oing beyond educational attainment, the authors w ­ ere able to construct a polygenic score for IQ specifically (for 257,841 individuals), self-­ reported math ability (for 564,698 individuals), and the highest level of math taken in school (for 430,445). Th ­ ese polygenic scores explain 7 percent to 10 percent of the variance. To summarize, the best state-­of-­the-­art research on ge­ne­tics with over one million p­ eople of Eu­ro­pean ancestry shows that genes directly explain 11 ­percent to 13 ­percent of the variation in educational attainment and a similar amount of variation in IQ. The team of scientists conducting this research formed the Social Science Ge­ne­tic Association Consortium and are utilizing best practices in science—­ data sharing, complete transparency in methods, replication studies—as they continue to publish results in the world’s top science journals. Their website summarizes the results of their work: Educational attainment is primarily determined by environmental ­factors, not genes. . . . ​The bottom line is that individual ge­ne­tic variants

144 • Chapter 5

have very ­little explanatory power for educational attainment, and even composite indexes of millions of ge­ne­tic variants have too ­little explanatory power to usefully predict any individual’s educational attainment.76 This is a scientific revolution, but not one that many psychologists expected. Between 1994 and 2018, scientists have gone from suggesting that IQ and educational attainment are mostly ge­ne­tic and that genes account for 40 ­percent to 80 ­percent of variation in a population to saying that genes have “­little explanatory power.” This is quite a reversal. To be fair, the Genome-­ Wide Association Studies methods d­ on’t capture all rare SNPs or complex interactions; ­f uture research ­will likely boost the amount of variation explained by polygenic scores. Still, even with twin studies, the best estimate for the heritability of education for a given cohort likely lies somewhere between 13 ­percent and 40 ­percent, but not 40 ­percent to 80 ­percent. Furthermore, even ­these very rigorous genomic-based studies are biased in ­favor of finding ge­ne­tic effects. As Augustine Kong and colleagues have found, genes that parents do not transmit to their ­children still appear to have causal effects on educational attainment. They dub this phenomenon “ge­ne­tic nurture” and estimate that it explains 30 ­percent of observed heritability.77 Other research also finds that a m ­ other’s polygenic score for education is significantly correlated with her child’s educational attainment, even ­after controlling for her child’s genes.78 Thus, what­ever the preferred estimate from the 13 ­percent to 40 ­percent range may be, it needs to be reduced to something like 9 ­percent to 28 ­percent. At this point, t­ here is no evidence that t­ here are meaningful differences between ancestral groups (or races) with re­spect to the frequency of genes that predict education or intelligence. Thus far, most of the ge­ne­tic samples analyzed—­including in the largest studies described above—­have come from ­people of Eu­ro­pean ancestry. Thus, the correlations between education and SNPs are ­limited to the environments of ­people from Eu­rope. The most recent and robust analy­sis from James Lee and his collaborators found that their polygenic score did not work for a small sample (1,519) of African Americans in that it explained only 1.6 ­percent of the variation in their educational attainment. This suggests that genes commonly found in highly educated white populations may be just as likely to be found in African Americans with low

Unequal Access to Education  •  145

levels of education.79 Scientists still have far to go before understanding the ge­ne­tics of education, particularly across cohorts and ancestral groups. Meanwhile, even within ­people of Eu­ro­pean ancestry, ­there is l­ittle to suggest that social class has any ge­ne­tic origins. One U.S. study finds that high-­income f­ athers have c­ hildren with slightly higher polygenic scores for education, but paternal income explains only 1.2 ­percent of the variation.80 Using a larger number of samples—­including data from the UK and New Zealand—­and an improved method for estimating polygenic scores, another research team did find a more substantial link between parental socioeconomic status and the polygenic scores of their ­children.81 Still, the correlations ­were weak to moderate. Almost all of the variation in ge­ne­tic advantage (in terms of higher education)—­over 90 ­percent—is not accounted for by parental socioeconomic status.82 This suggests that if you think you can judge a child’s ge­ne­tic propensity to be intelligent or highly educated on the basis of the status of their ­family, you’d be wrong. As Plato wrote: “As all are of the same original stock, a golden parent w ­ ill sometimes have a silver son, or a silver parent a golden son.”83 The bottom line is that the latest ge­ne­tics research supports the notion that highly intelligent ­people can come from any class or ethnic group. Having parents with genes advantageous to intelligence or education gives one a small advantage, not a big one, for two reasons: The environment ­matters a ­great deal, and the genes associated with intelligence are extremely common. Scientists have found no relationship between very high IQ and the presence of rare ge­ne­tic alleles.84 Likewise, none of the specific genes identified as significant predictors of IQ qualifies as rare in the latest genome-­wide studies with 1.1 million ­people.85 The average SNP that predicts intelligence is pre­sent in half of all ­people with Eu­ro­pean origins. The genes that ­were strongly associated with intelligence ­were no less common than genes that did not predict intelligence. Thus, the odds are that every­one has at least some of the genes that predict high IQ. Highly intelligent p­ eople may be somewhat more likely to have a beneficial combination of many common genes, but they score well and achieve higher levels of education largely based on environmental advantages. To summarize, the most scientifically rigorous non-­twin-­based estimates suggest that as much as 13 ­percent of differences between individuals in educational attainment is the result of genes for a given population, and even twin studies—­despite their upward bias in estimating heritability—­a lso suggest

146 • Chapter 5

that most of the variation in education (60  ­percent) is explained by the environment. The key caveat is within a given population. From twin studies to genomic-­ based polygenic scores, all the estimates of heritability are l­imited to a specific population—­that means p­ eople born at roughly the same time in the same place, typically of the same race. The large-­scale genomic work described above used only ­people of Eu­ro­pean ancestry, for example, ­because the sample sizes are too small for p­ eople of recent African descent. The implication is that heritability estimates do not apply between generations and groups of ­people. It is a very dif­fer­ent to ask: What percentage of the variation in IQ between all of humanity is the result of genes? Right now, no one can give a good answer, but we can say that it is much lower than the current estimate for heritability in IQ or educational attainment for several reasons. First, even genomic-­based heritability estimates are inflated ­because they ignore the purely environmental sources of variation between generations. If genomic scholars could include every­one born between 1880 and 1980 in their database, and if they did not control for cohort effects, they would find that the explanatory power of genes shrinks dramatically as a result of economic development (or the Flynn effect). Second, as shown above, we know that IQ differences between countries are at least largely environmental and likely 100 ­percent environmental. Thus, when the same relatively uneducated parents raise their c­ hildren in northern Eu­rope before age ten instead of in a poor country ­until age ten, IQs go up by about 10 points. The subsequent generation of c­ hildren born in northern Eu­rope w ­ ill no doubt benefit from the fact their parents w ­ ill be raising them in a more literate ­house­hold, with all the other advantages that go along with living in a rich country with high-­quality public ser­vices, schools, healthcare, and well-­functioning markets. We know that sustained economic development has coincided with growth in IQ by roughly 0.3 points per year, so it stands to reason that economic development in currently poor countries would raise IQ by a similar magnitude and has already started to do so. As it happens, most poor countries ­today—­those not classified by the World Bank as high-­income—­are even poorer than the United States was 50 year ago, when IQ scores for the United States ­were approximately 15 points lower.86 So, if one considers all h ­ umans who ever lived, the ge­ne­tic basis of intelligence—as mea­sured by IQ tests—­

Unequal Access to Education  •  147

would be infinitesimally small. The share of h ­ uman variation explained by ge­ne­tics is negligible, b­ ecause environmental differences are absolutely dominant. Thirty years ago, scientists thought IQ was dominated by genes. Con­ temporary social science shows that genes are impor­tant in explaining variation in education and other social outcomes, but the environment is more power­ful, and the two interact in complex ways that scientists are only beginning to understand. The under­lying capacity of individuals to perform cognitively demanding tasks is far more egalitarian than twentieth-­century science led us to believe. The next chapter uses historical analy­sis to illustrate how group differences in intelligence—­which have reversed rankings in very recent years—­cannot be explained by genes. Instead, po­liti­cal power and culture—in the form of readily transferable f­ amily norms—­explain the gaps.

6 The Historical Contingencies of Group Differences in Skills

Writing in the fourth c ­ entury BCE, at a time when slavery was widespread, Aristotle argued that some ­people are by nature slaves. He distinguished between p­ eople who become enslaved ­because they lose a war and ­people who are naturally servile and worthy only of being slaves. He argued that w ­ omen and certain classes of p­ eople are fundamentally “inferior” and should justly be ruled by their superiors. A ­ fter developing his concept of natu­ral slavery, Aristotle l­ ater attributed the trait to dif­fer­ent races, ascribing a tolerance for tyrannical rule to non-­Hellenes (non-­Greeks): “Barbarians, being more servile in character than Hellenes, and Asiatics than Eu­ro­pe­ans, do not rebel against a despotic government. Such royalties have the nature of tyrannies b­ ecause the p­ eople are by nature slaves.”1 It is worth noting that this was not the prevailing opinion in his day. Indeed, his own teacher, Plato, rejected the notion of natu­ral slavery and Aristotle’s view of ­women. Aristotle acknowledged as much: “­Others affirm that the rule of a master over slaves is contrary to nature, and that the distinction between slave and freeman exists by law only, and not by nature; and being an interference with nature is therefore unjust.”2 The Aristotelian theory of the natu­ral slave has endured a long time. White po­liti­cal leaders of the U.S. South ­were explicit in their view that black ­people ­were naturally inferior and thus deserving of slavery. Alexander H. Stevens, vice president of the Confederacy, criticized Thomas Jefferson for espousing the view that slavery ­violated the laws of nature. ­These ideas ­were wrong, according to Stevens, ­because they rested upon the assumption of the equality of races. This was an error. . . . ​Our new government is founded upon exactly the opposite

Group Differences in Skills  •  149

idea, its foundation are laid, its corner-­stone rests upon the ­great truth that the Negro is not equal to the white man. That slavery—subordination to the superior race—is his natu­ral and normal condition.”3 The twentieth-­century version of this theory gave it an air of spurious scientific legitimacy by linking it to ge­ne­tic research, but as we saw in the last chapter and w ­ ill explore further ­here, ­those assumptions are fundamentally flawed. While ­there have been many efforts throughout history to attribute social status to some essential group trait, the theories make implausible scientific and evolutionary assumptions and fail to acknowledge or account for the fact that the social status of ­actual groups fluctuates and reorders itself in ways that essential traits cannot explain.

WHEN RACISM BECAME AN EPIDEMIC

The pro­cesses of urbanization, development of local po­liti­cal autonomy, and then industrialization that lifted northern Eu­rope from its long-­standing global mediocrity to economic and military power coincided with a deeper understanding of the role of evolution in long-­term ­human development. Many assumed that the f­ actors at work in the transformation of primitive apes to Homo sapiens could also be seen in the differences between races, and like Aristotle, ­those of northern Eu­ro­pean stock often mistook their very recent success as being the result of essential traits that they believed distinguished them from other p­ eople, rather than the result of exogenous forces. In the first few de­cades of the twentieth c­ entury, the near consensus among U.S. scholars and scientists—if not the general public—­was that ge­ne­tic endowments varied between races, that IQ was a broad mea­sure of h ­ uman competence and almost entirely ge­ne­tic, that northern Eu­ro­pe­ans had the highest IQ, and that therefore northern Eu­ro­pean genes ­were at risk of dilution from other ­people—­unless only ­those non–­northern Eu­ro­pe­ans with the highest IQ w ­ ere allowed into the United States. The history of t­ hese ideas has been documented extensively by the excellent work of historians like Nell Irvin Painter and Louis Menand, who point out that many scientists believed the races evolved from fundamentally distinct populations, if not species.4 Modern ge­ne­tic research has since proven that all h ­ umans share a common origin in Sub-­Saharan Africa.

150 • Chapter 6

Flawed science—in combination with widespread racism—­led to a vigorous eugenics movement, as well as immigration policies designed to discriminate against p­ eople who w ­ ere not from northern or western Eu­rope. The eugenics movement resulted in the forced sterilization of tens of thousands of Americans during the first six de­cades of the twentieth ­century; 63,678 sterilization cases w ­ ere documented between 1907 and 1964.5 Just over half of ­those sterilized ­were deemed mentally deficient, meaning they registered a low score on an IQ test. The other half ­were sterilized mostly ­because they ­were diagnosed as mentally ill. Moreover, many c­ hildren—­disproportionately from immigrant and black families—­were sent to institutions for the mentally deficient, where they ­were put to work d­ oing routine jobs that required low levels of cognitive ability. An IQ score below 80, as tested by a local public school or physician, put a child at risk of forced institutionalization u­ nder the New York State M ­ ental Deficiency Law, which was passed in 1919. In 1920, t­ here ­were roughly 6,000 ­children—­deemed “defectives” by the state of New York—­who w ­ ere removed from their families and placed in institutions.6 The state estimated 40,000 “defectives” living outside of institutions. The most common cause listed for their admission was “hereditary.” Conditions ­were such that 96 ­children died while living in ­these institutions from 1919 to 1920.7 In other words, ­children raised in poor or uneducated families ­were at high risk of low test scores and being deemed genet­ically defective and denied basic freedoms, with no effort made to remediate gaps in their education.

THE CON­T EMPORARY HEREDITARIAN VIEW OF INTELLIGENCE

The idea that ­people can be classified into distinct races and ordered as part of a hierarchy of races was the dominant intellectual view in the early twentieth c­ entury. White scholars of “Anglo-­Saxon” or northern Eu­ro­pean ancestry openly wrote in leading academic journals that other groups ­were inferior intellectually and socially as a result of their ge­ne­tic endowment, and they used this logic to motivate eugenics policies and regulations on immigration and wages.8 It should be said that IQ testing was not usually involved in t­ hese debates. The IQ lit­er­a­ture from the period shows that the tests ­were conducted for immediate practical considerations—­such as assigning soldiers to par­tic­u ­lar

Group Differences in Skills  •  151

occupations or flagging c­ hildren for educational remediation or access to advanced courses—­and the authors often mentioned pos­si­ble environmental explanations, even as they seemed to almost universally appeal to ge­ne­ tics in explaining group differences. Starting in the second half of the twentieth ­century and into the twenty-­ first, this theory of racial hierarchy has been largely decimated by social science research and cultural and historic trends. Still, as I discussed in chapter 1, the racist views of a minority of citizens and politicians still carry po­liti­cal weight—­and arguably affect impor­tant public policy positions on issues such as immigration and education. Prominent cultural and academic figures still maintain the basic tenets of racial hierarchy, which are openly espoused on social media, even if rarely in academic journals or books. The fact that t­ hese views are currently unpop­u­lar among Hollywood celebrities, athletes, and entertainers makes them attractive to a certain segment of the population, as I discussed in chapter 1, when describing the “Alt-­R ight.” ­Those who defend the theory of racial hierarchy portray themselves as stalwart defenders of truth in the face of po­liti­cally correct mobs.9 In psy­chol­ogy lit­er­a­ture, the theory that ­there are impor­tant racial differences in the genes that determine success in life is now called hereditarianism. In its essence, it attributes differences in social status between groups of ­people—­including educational attainment and cognitive per­for­mance in tests—to innate ge­ne­tic differences. This belief that group differences between ­people whose ancestors came from dif­fer­ent continents are innate is the textbook definition of racism w ­ hether or not any animosity is evinced ­toward lower-­status groups.10 ­These views are also stridently classist in that they attribute low educational attainment, low test scores, single-­parent child-­rearing, and crime, as well as other difficulties confronting the working poor, ­whether they are white or other­wise, to ge­ne­tics, rather than social and po­liti­cal in­equality. It would be tempting to reject such views solely ­because they are racist or classist, but that would be a ­mistake po­liti­cally and intellectually. Po­liti­cally, rejecting unpop­u­lar views without considering their foundations creates a backlash and subjects well-­meaning critics to charges of sentimentality or wishful thinking, while giving hereditarians the status of “realists.” It’s no way to win an argument. Hereditarian views should be subject to scrutiny and testing just as any other theory in science and should be treated on ­those terms. In fact, hereditarianism can be shown to fail using empirical analy­sis and logic.

152 • Chapter 6

Two of the intellectual leaders of the modern hereditarian movement ­were the late Arthur Jensen and Phillipe Rushton; both died in 2012. Jensen was a hugely influential but controversial figure in psy­chol­ogy, who had a long ­career as a professor at the University of California, Berkeley, and published many papers that seemingly had l­ittle to do with race. His colleague Rushton taught psy­chol­ogy at the University of Western Ontario. Like scholars of the early twentieth ­century, Jensen and Rushton observed differences in social status between groups and concluded that ­these were due to “the natu­ral order.” Rushton devoted much of his c­ areer to mea­sur­ing and discussing differences in cognitive per­for­mance between individuals and groups and arguing for their ge­ne­tic origin, and Jensen seems to have held the view for de­cades at least. In 2005, they published a defense of hereditarianism, based on their review of the psy­chol­ogy lit­er­a­ture. Similar themes came up in a widely cited and criticized 1969 article by Jensen claiming that IQ cannot be increased by postnatal environmental influences—an absurd and false claim, as chapter 5 in this book proves.11 The fundamental reasoning of hereditarians is quite ­simple. According to their interpretation of the psy­chol­ogy lit­er­a­ture, genes explain roughly half of the variation in individual per­for­mance on cognitive exams—­which include IQ tests. Group differences, Jensen, Rushton, and hereditarians reason, can be thought of similarly. They are enormously impressed by the observation that the mean test scores of African Americans—on vari­ous cognitive exams—­are consistently lower than t­ hose of white Americans (by roughly one standard deviation, in statistical terms, or 15 IQ points). They also note that Asians and Jews of Eu­ro­pean descent (the Ashkenazi) living outside of Israel tend to score higher than non-­Jewish white ­people. They note that studies have found that mixed-­race black-­white c­ hildren score in between ­children with two white parents and c­ hildren with two black parents. Further, they cite research showing that ­people living in Sub-­Saharan Africa score even lower than African Americans and believe this provides further evidence that racial differences are ge­ne­tic in origin, since many African Americans have white ancestors as a result of the raping of enslaved w ­ omen.12 The IQ tests are not biased, they argue, ­because in each of ­these groups, ­people who score better on cognitive exams exhibit greater success, as mea­sured by educational attainment, income, or other outcomes. Richard Herrnstein and Charles Murray made some of t­ hese points in their controversial 1994 book, The Bell Curve, which pop­u­lar­ized the ­earlier

Group Differences in Skills  •  153

research of Jensen, Rushton, and other hereditarians, while contributing their own novel analy­sis and interpretation of the lit­er­a­ture.13 The Bell Curve is well written, well or­ga­nized, and extremely useful for understanding how a segment of academic psychologists and social scientists understand differences in social status across race and class. I d­ on’t think it would be fair to say that Murray and Herrnstein misunderstand or grossly distort the lit­er­a­ture on IQ; rather, the book is fundamentally wrong, as I see it, b­ ecause they accept so much of the scholarship uncritically and with insufficient attention to contradictory evidence and interpretations, especially from other disciplines. In many ways, Herrnstein and Murray’s interpretation of the evidence was typical, at least among older white scholars who specialize in intelligence research. They point to survey data of prominent academics, which suggest that just u­ nder half believe group differences in IQ are partly ge­ne­tic in origin. ­A fter The Bell Curve came out, a group of 52 of the 100 most prominent researchers on intelligence published a letter in the Wall Street Journal, which endorsed many of Murray and Herrnstein’s claims, such as the claim that IQ heritability is between 40 ­percent and 80 ­percent. This task force endorsed the following statement: “Most experts believe that environment is impor­ tant in pushing the bell curves apart, but that ge­ne­tics could be involved too.” Murray and Herrnstein could be characterized as emphasizing the importance of ge­ne­tics to a greater extent than the average expert in the field, but many w ­ ere open to at least some role for ge­ne­tics in explaining group differences.14 Along t­ hese lines, a survey conducted in 2013 and 2014 of scholars who have published in journals specializing in intelligence research confirmed that many view genes as impor­tant in explaining group differences in IQ performance—­between nations, races, and ethnic groups. The average respondent attributed one-­sixth to one-­fi fth of differences to genes.15 Disproportionate Group Achievement

The hereditarian community has put a g­ reat deal of stock in the argument that Jewish and Asian intellectual achievements should be regarded as evidence that group differences are explained by genes, not the environment. While hereditarians concede that severe discrimination against African Americans is an environmental force that has lowered scholastic achievements, they suggest that white Christian Eu­ro­pe­ans are not discriminated against and so the higher per­for­mance of Eu­ro­pean Jews and Asians must not

154 • Chapter 6

be caused by environmental f­actors. In fact, Jews and Asians have been subjected to intense discrimination in the United States. They reason that if ge­ne­tic ­factors cause Asians and Jews to demonstrate high achievement, ge­ ne­tic ­factors likely explain most of the differences across all groups. Thus, ­these arguments have impor­tant implications, and for that reason, and ­because I believe the arguments to be entirely unsound, it is very impor­tant to understand the details. The average American Jew and Asian American is considerably more educated and affluent than the average non-­Hispanic, non-­Jewish white person, black person, Native American, or Hispanic. Gallup data from 2015 to 2017 show that 43 ­percent of Jews and 35 ­percent of Asians aged 25 to 50 have earned a gradu­ate degree, compared to 16 ­percent of non-­Jewish white ­people. Jews and Asians are also far more likely than white p­ eople to be in a h ­ ouse­hold with at least $240,000 in income (figure 6.1). In 2015, Asian American 12th-­grade students outscored their white peers on a national math assessment by 0.3 standard deviations, whereas scores on reading and science w ­ ere roughly equal.16 More significantly, from 1986 to 2015, Asian Americans outscored white p­ eople on the math section of the Standard Achievement Test (SAT) for college entrance by a comfortable margin, including by as much as 0.6 standard deviations in 2015. White ­people consistently outscored Asians over that period on reading, but on writing, Asians closed the gap and outscored white p­ eople a­ fter 2008.17 The high academic per­for­mance of many Asian Americans during high school has increased their share of students at top American colleges even though Asian ­A mericans need to score higher on average than other groups to gain admission, at least to Harvard.18 Likewise, Jewish Americans once faced similar discriminatory quotas at prestigious colleges, even as they made outstanding contributions to the nation’s academic achievements. From 1901 to 2018, p­ eople with at least one Jewish parent or grandparent have won 22 ­percent of all Nobel Prizes, including 37  ­percent of the awards granted to Americans. Jews also have won 38 ­percent of U.S. Medals of Science. Most of the Jewish Nobel winners have been American and ­were awarded the prize in the second half of the twentieth ­century.19 Of course, Jews have also made valuable and celebrated contributions across all fields, including the arts, with, for example, many commercially successful and Acad­emy Award–­winning directors including Steven Spielberg, Woody Allen, and Stanley Kubrick.

Group Differences in Skills  •  155 43 35

16 7

3

3

Hispanic

16

11

11

Native American

10 2

Black

5

White

Asian

Jewish

Percentage with a graduate degree Percentage with household income of $240,000 Figure 6.1. ​­Percent of U.S. residents aged 25–50 with a post-­bachelor’s degree and ­house­hold income of $240,000 or higher, 2015–2017 Source: Gallup Daily Tracker phone survey of 314,101 randomly selected U.S. residents aged 25 to 50.

A number of social scientists—­Gregory Cochran, Jason Hardy, Henry Harpending, Richard Lynn, Charles Murray, Nicholas Wade, and Steven Pinker—have published academic or popu­lar commentaries noting ­these achievements and embracing, with some reservations, the theory that ge­ne­ tic advantages explain this success.20 In considering how extraordinary the contributions mentioned above are, some of ­these writers refer to the tiny proportion of Jews in the global population. That is a misleading comparison. The majority of the global population was illiterate for the first half of the twentieth c­ entury, and as late as 1975, only 3 ­percent of the global population had any postsecondary education.21 In 2015, that share had increased, but only to 10 ­percent.22 Lynn reports that Jews are ten times more likely than non-­Jews in the United States to have won the Nobel Prize, which is accurate considering that Jewish p­ eople comprised around 4 ­percent of the U.S. population around the ­middle of the twentieth ­century and have won 37  ­percent of Nobel Prizes.23 Yet, Jews ­were so disproportionately educated that in 1970 Jews comprised 23.5  ­percent of the U.S. population with at least six years of higher education, including 30 ­percent of the population aged 25 to 29.24 In 1970, 18.2 ­percent of the Jewish population had obtained at least a master’s

156 • Chapter 6

degree compared to 2.7 ­percent of the total American population. Since a gradu­ate education is effectively a prerequisite for a scientific award, the disproportionate number of leading scientists is not so far out of proportion from the Jewish share of the population with the required level of education. Still, this begs the question as to why Jews are so much more likely to obtain high levels of education. Hereditarians have advanced two theories. The most comprehensive ge­ne­tic theory was developed by Cochran, Hardy, and Harpending. Noting that ancient Jews did not distinguish themselves in any way that would indicate higher intelligence relative to other populations, they point to the unique occupational structure of Jews in professional and banking jobs in Eu­rope, which they believe gave Jews a unique opportunity to profit from cognitive ability. Cochran and his co-­authors argue that merely 500 years of selective breeding are necessary to transform a group with ordinary intelligence—­ancient and medieval Jews—­into a group with extreme intelligence (a standard deviation above the mean—an IQ 115). They get t­ here by assuming that IQ is 80 ­percent hereditary within a given population and that Jewish overrepre­sen­ta­tion in highly skilled occupations gave them unique opportunities to afford more c­ hildren. In other words, intelligence was rewarded among Jews with extra ­children, but not among any other group to the same extent, and they speculate (with no factual basis) that as a result of this pro­cess, the average Jewish child could have been born to a ­family with one extra IQ point above the population. ­These two assumptions—­ heritability of .8 and selective breeding of one extra IQ point—­allow them to estimate 500 years as necessary for Jews to go from average to extra­ordinary. Using the sample assumptions but changing IQ heritability to .25 (which would be a high-­end estimate from molecular biology) requires 1,500 years. Without selective breeding, the theory does not work at all. The second theory explaining Jewish achievement has been developed by the economists Maristella Botticini and Zvi Eckstein.25 This is not a hereditarian theory at all, but it has subsequently been adapted by Wade and Murray, as I w ­ ill discuss. Botticini and Eckstein point out that ancient Jews ­were nearly all illiterate farmers during the time of the Roman Empire and up through the birth of Christ. Breaking from previous scholars, Botticini and Eckstein document that ­there ­were no ­legal restrictions prohibiting Jews from owning land or working in any occupations, even in the Roman, Byzantine, Persian, and Islamic Empires, and yet Jews in the medieval era ­were disproportionately engaged in business owner­ship, trading, and banking. The

Group Differences in Skills  •  157

s­ imple explanation is that Jews w ­ ere far more literate than e­ very other large community, and so they found employment in jobs that benefited from literacy. The reason for high rates of ancient Jewish literacy is no mystery. Around the year 63, a Jewish high priest named Joshua ben Gamla issued a law requiring e­ very Jewish f­ather to educate his son starting around age six. According to Botticini and Eckstein, this law was prompted by the traumatic destruction of the Second Jewish T ­ emple in Jerusalem by the Roman Army ­after the Jews attempted to revolt against Roman colonization. By diffusing literacy, Jewish leaders hoped to safeguard and preserve their laws and teachings.26 As mentioned in chapter 2, universal childhood education was not a new idea: Plato had actively recommended it centuries before, even for girls, but evidence from the ancient world suggests that none of the large imperial powers invested in public primary education.27 In Plato’s Athens, ­there is some evidence for widespread but by no means universal literacy among craftsmen and o­ thers in high-­paying occupations and jurists w ­ ere expected to be literate. Schools w ­ ere common, and even slaves w ­ ere often taught by their masters. One source, Diodorus Siculus, claims that publicly funded education was the law in the Sicilian city of Catania.28 But ­because t­ hese literacy-­promoting norms w ­ ere tied to city-­states, they could be eliminated by raiders or larger powers waging successful wars with ­those city-­states. For Jews, by contrast, t­ hose norms w ­ ere tied to their religious practice, not their status as citizens. Wherever Jews lived, they set up synagogues that acted as primary and even secondary schools. Local taxes funded the schooling, which allowed even lower-­income f­athers to educate their ­children. The Roman Jew Josephus wrote that “­children’s education is the principal care among the Jews.”29 Shortly a­ fter the requirements for literacy w ­ ere made, Botticini and Eckstein document a large decrease in the Jewish population that cannot be readily explained by death and seems to be the result of conversion. They argue that Jewish ­fathers who could not afford to educate their sons chose conversion over the social sanctions they faced as Jews for disobeying this impor­ tant norm. It is ­here that hereditarians such as Charles Murray pick up the argument as evidence that only the most naturally gifted Jews remained Jews and that the rest converted to Chris­tian­ity or other religions. Hence, according to

158 • Chapter 6

Murray and Wade, they argue Jewish achievement is explained by the fact that Jewish ­legal requirements pushed out the Jews with low IQ genes.30 To be clear, this argument is not made by Botticini and Eckstein. Evaluating Hereditarian Theories for Jewish Achievement

­There are three major flaws with the theories described above. First, according to the sorts of mea­sures used by hereditarians, Jewish intellectual achievements ­were similar to ­those of other groups u­ ntil the nineteenth c­ entury and certainly nowhere near the level of the mid- and late twentieth ­century, as Michael Goldfarb has pointed out.31 Jewish contributions to the Industrial Revolution also came rather late, with a few exceptions. This is hard to reconcile with any kind of ge­ne­tic explanation for more recent achievement ­because genes are very slow to change even ­under the most aggressive assumptions. Second, IQ has been extensively mea­sured for the Jewish population living in the United States and more recently for Israeli Jews. ­There is strong evidence that Jewish Americans had no more than average and even lower intellectual abilities in the first half of the twentieth ­century than the native white population, but the data do suggest rapid growth in IQ between generations as Jews became more educated. Israeli Jews, meanwhile, perform at roughly the global average for developed countries and well below residents of high-­scoring Eu­ro­pean and Asian countries. Third, t­ here is no evidence for selective breeding among American Jews that could explain the growth in achievement or IQ. The most elite American Jews—­measured in several ways, including IQ—­have fewer c­ hildren than lower-­status Jews, and this gap is even more pronounced among Jews than it is for non-­Jews. I ­will discuss each of ­these points in detail below. The focus on the Jewish population stems from the central role hereditarian theorists have given them. Some aspects of the discussion are applicable to explaining changes in the status of Asian and non-­Jewish northern Eu­ro­pean populations as well, at vari­ous points in history. The overarching theme from the study of group status—­ measured by education, income, or IQ—is its fluidity and high rate of turnover. No group of ­people has persisted in maintaining sustained levels of high status for long enough periods for population ge­ne­tics to explain their success. Likewise, between slavery and Jim Crow, African Americans living

Group Differences in Skills  •  159

in northern states achieved at intellectually higher rates than Eu­ro­pean and Asian immigrant groups and possibly even native white Americans. The reason is that the genes that ­favor productivity, intelligence, and conscientiousness are common to humanity, and economic status depends on po­liti­cal history and cultural practices that can change rapidly for the better or worse. Eu­ro­pean Jews before and during the Era of Mass Migration

­ ese hereditarian theories for Jewish achievement posit sustained long-­term Th advantages in cognitive ability ­going back hundreds of years. Cochran and his co-­authors provide anecdotal evidence that Eu­ro­pean Jews w ­ ere overrepresented in cognitively demanding professions such as finance. Richard Lynn, meanwhile, also claims that Jews are genet­ically advantaged with re­spect to intelligence but does not offer any explanation as to when or how this extra intelligence may have evolved in the Jewish population.32 He does go to ­great lengths to assem­ble facts that he believes support his argument, and so I w ­ ill address them. However, a point-­by-­point rebuttal would be extremely tedious, given that I found flaws with his analy­sis in almost ­every case, but I ­will attempt to summarize the crucial points. One type of flaw in Lynn’s analy­sis is that he often shows that a high percentage of Eu­ro­pean Jews worked in professions in a par­tic­u­lar town or city, but ignores the fact that selective migration accounts for much or even all of this. For example, in the Rus­sian Empire, only Jews in elite professions w ­ ere allowed to leave the Pale of Settlement;33 thus, Jews living outside the Pale of Settlement w ­ ere by definition from elite families. It is nonsensical to then suggest, as Lynn does, that the high percentage of professional Jews in St. Petersburg is representative of the larger Jewish population in Rus­sia. In fact, it was well documented by the Rus­sian government that the Jewish population within Poland and the Pale of Settlement suffered from high rates of poverty and worked mostly in low-­paying occupations.34 In 1913, two-­thirds of Eu­ro­pean Jews—­who themselves comprised most of the global Jewish population—­lived in the Rus­sian Empire or Romania, with another 21 ­percent in Austria-­Hungary.35 According to Bernard Wasserstein, a leading historian of the Ashkenazi, in the early twentieth c­ entury, “The g­ reat mass of Jews in east-­central Eu­rope w ­ ere sunk in dire poverty—­ and sinking further into total immiseration.”36 Outside of Rus­sia, few Jews ­were farmers, but neither did they occupy highly skilled occupations in large numbers. In Prus­sia, in 1849, 48 ­percent

160 • Chapter 6

of Jews worked in commerce—­they w ­ ere mostly retail merchants, peddlers, and dealers, as listed by the census. While more reliant on literacy, perhaps, it is not clear that ­these jobs are more cognitively demanding than farming, which is also a small business enterprise that requires knowledge of production techniques, agricultural science, supply chains, and distribution networks. Only 1 ­percent of Jews worked in banking and 2 ­percent as physicians or teachers. A large portion (20 ­percent) w ­ ere also mechanics or handicraftsmen, while 10 ­percent w ­ ere servants, and 10 ­percent w ­ ere paupers. In 1861, again in Prus­sia, 57 ­percent worked in commerce, but still only 1 ­percent in banking and 3 ­percent as physicians or teachers. In both periods, only 1 ­percent w ­ ere farmers. Statistics w ­ ere similar for Italy and Budapest.37 De­cades ­later, in Germany, the status of Jews appears to have been much better, and Jews ­were far more likely than the general population to be business leaders, scientists, and mathematicians, up u­ ntil their expulsion and persecution by the Nazi government in the early 1930s.38 Incredibly, German Jews comprised roughly 16 ­percent of corporate board members, despite constituting only 1 ­percent of the population.39 Still, German Jews ­were hardly representative of the global or even Eu­ro­ pean Jewish population. They constituted 6 ­percent of Eu­ro­pean Jews.40 Moreover, the extent to which this population can be considered to have unique ge­ne­tic characteristics is highly questionable. In Berlin, in 1899, 18 ­percent of Jews married Christians, and it is reported that more prosperous Jews ­were more likely to do so.41 Nazis ­later persecuted ­people as Jews if even one grandparent was Jewish. American Asians, Jews, and Black P ­ eople in the Late Nineteenth and Early Twentieth Centuries

Data from U.S. sources corroborate the un-­extraordinary social status of American Jews and Asians in the early twentieth c­ entury. My analy­sis of U.S. census data from 1920 shows that men aged 30 to 40 who listed Yiddish or Hebrew as their ­mother’s tongue (which would have been the vast majority of Jews) w ­ ere less likely than native white p­ eople to be working in professional occupations (2.7 ­percent compared to 3.6 ­percent). White ­people born in the northeastern United States w ­ ere twice as likely as Jews to work in the professions. ­These included jobs such as engineers, academics, doctors, and l­awyers. Only 1.3 ­percent of Asians worked as professionals at the time. The small share of Jews working in professional occu-

Group Differences in Skills  •  161

pations is even more striking when one considers that only 1 ­percent of Jews worked as farmers at a time when 25 ­percent of native white p­ eople ­were farmers. Home ownership rates ­were also low for Asians and Jews, even compared to other immigrants, perhaps ­because they lived in urban areas with higher home prices. Literacy in any language was much higher for Jews—­and to some extent Asians—­than it was for Hispanic and Italian immigrants, but both Jews and Asians had lower literacy rates than the native white population and lower rates than northern Eu­ro­pean immigrants. Jews and Asians did stand out as being much more likely than other immigrant groups to be business o­ wners and man­ag­ers, but the share of o­ wners who actually employed other ­people was roughly equal to the same share for native white ­people. Where Asians and particularly Jews w ­ ere most distinctive was in their propensity to work as self-­employed man­ag­ers—­such as in­ de­pen­dent proprietors (­table 6.1). Other data from the period suggests that Jews w ­ ere disproportionately likely to be working in finance, but ­these jobs ­were still dominated by Anglo-­ Saxons throughout the early twentieth ­century, and the relationship appears to have been geographic. For example, 65 ­percent of traders on the New York Stock Exchange w ­ ere Anglo-­Saxon in 1910 and just 3 ­percent Jewish. By 1930, the number of Jews had risen to 8.6 ­percent, but Anglo-­Saxons still comprised 60 ­percent.42 But then 8.6 ­percent is not a large percentage considering that while Jews comprised roughly 4 ­percent of the U.S. population in 1929, they constituted 30 ­percent of the New York City population.43 Thus, from the perspective of the local population base, Jews w ­ ere underrepresented at the stock exchange. Likewise, while Jews ­were disproportionately chief executives of the largest companies in Germany, the same was not true in the United States. Just 4.6 ­percent of CEOs of the 200 largest U.S. companies in 1917 ­were Jewish, which roughly equaled their share of the population.44 Two-­thirds w ­ ere Episcopalian or Presbyterian, though they represented just 4 ­percent of the U.S. population.45 ­There is also no evidence to suggest that Jews or Asians ­were prominent in U.S. invention or science in the late nineteenth and early twentieth centuries—at least ­until the Nazis forced German Jewish scientists to leave in the 1930s.46 The 1920 census data, for example, shows that compared with U.S. white ­people, as well as immigrants from ­England, Germany, and

7.1% 5.8% 6.7% 20.8% 6.3% 1.4% 3.9% 3.7% 9.7% 0.9% 2.5% 5.7%

4.6%

4.2% 3.6% 2.7%

2.5% 1.6% 1.6% 1.5% 1.3% 1.2% 1.0%

0.9%

Share in managerial occupations

4.4%

15.5% 3.2% 3.6% 16.4% 26.3% 35.7% 22.7%

5.3% 26.3% 0.7%

9.1%

Share who are farmers

1.5%

4.9% 0.8% 3.2% 5.6% 0.4% 0.4% 0.7%

5.7% 2.6% 1.0%

4.1%

Share who are machinists

2.5%

6.4% 0.6% 1.3% 5.4% 8.5% 5.4% 2.0%

2.8% 7.8% 8.0%

4.1%

Share who own a business

44.0%

27.8% 49.1% 39.3% 30.3% 28.0% 37.3% 46.2%

29.3% 25.4% 24.8%

29.9%

Share who are laborers

74.2%

98.1% 94.6% 99.0% 98.6% 89.2% 77.9% 65.6%

99.6% 97.7% 92.2%

99.2%

Literacy

28.2%

43.0% 12.6% 28.0% 45.4% 10.0% 20.4% 15.5%

34.5% 44.6% 19.4%

37.4%

Home owner­ship rate

Source: 1920 complete count U.S. Census, with 9 million U.S. male residents aged 30–40. Steven Ruggles, Katie Genadek, Ronald Goeken, Josiah Grover, and Matthew Sobek. Integrated Public Use Microdata Series: Version 7.0 [dataset]. Minneapolis, MN: University of Minnesota, 2017.

Northeastern white, U.S.-born Born in ­England White, U.S.-born Yiddish-­speaking ­mother Born in Germany Northeastern black Born in Ireland Born in Sweden Asian Black Spanish-­speaking ­mother Italian-­speaking ­mother

Share in professional occupations

­ able 6.1.  Men aged 30 to 40 living in the United States in 1920, by occupational category, literacy, home T owner­ship status, and ethnic background

Group Differences in Skills  •  163

Sweden, Jews and Asians ­were greatly underrepresented as scientists, machinists, and engineers, the occupations most likely to initiate a patent.47 This puzzle has been noted by social scientists who have delved deeply into Jewish social status and achievement. It is worth quoting Joel Mokyr, a noted Israeli American economic historian, at length: By 1850, technological innovation had become a central source of economic change. Yet despite their huge head start eco­nom­ically, Jews played an almost imperceptible role in the history of science and technology before and during the early Industrial Revolution. . . . ​Given the high investment in ­human capital by Jews, their obsession with learning and books, and their devotion to the education of their sons (many of whom mastered high-­skill crafts), one might have predicted other­ wise. . . . ​Yet in the annals of the Industrial Revolution, Jews are hard to find.48 Biographies of prominent inventors during the most prolific period of the Industrial Revolution show close to zero Jews or Asians—­either in Eu­rope or the United States. For example, the National Inventors Hall of Fame aims to recognize the greatest inventions ever patented at the U.S. Patent and Trademark Office (USPTO). Of the 104 inventors born between 1743 and 1850—­which include renowned inventors such as Thomas Edison, Alexander Graham Bell, Cyrus McCormack, and Eli Whitney—­I could not identify any who w ­ ere of Jewish or Asian heritage. Thereafter, only a handful are apparently Jewish ­until the twentieth ­century; among the early inventors are Eli Berliner, who was born to German Jewish merchants in 1851 and contributed to the gramophone and microphone, and Willem Einthoven, who was born to a gentile m ­ other and a Dutch Jewish f­ ather in 1860 and patented the electrocardiograph, which also won him the Nobel Prize for medicine in 1924.49 It is well known that northern Eu­ro­pean gentiles—­particularly the British ­people—­were the driving force of the Industrial Revolution. This is difficult to explain if, as hereditarians argue, the ­great contributions of Jews to science and invention l­ater in history was a result of their ge­ne­tic advantage for intelligence over northern Eu­ro­pe­ans and t­ hose of African origin. Of course, northern Eu­ro­pean gentiles had greater po­liti­cal power and autonomy than Eu­ro­pean Jews, and even hereditarians concede that environmental ­factors can make a difference. And yet, the hereditarian claim that

164 • Chapter 6

recent achievements by Jews and Asians ­were based on long-­standing ge­ne­tic advantages collapses when comparing their per­for­mance to African Americans during the Industrial Revolution. African Americans suffered from po­ liti­cal oppression at least as much as Jews and Asians, and despite this, the sons and ­daughters of escaped or emancipated African American slaves arguably made greater contributions to the Industrial Revolution than Jews or Asians. It is not as improbable as it may seem: Black ­people in the Northeast had higher literacy rates than e­ ither Asians or Jews in 1920, according to census data. It is well documented that African Americans made impor­tant contributions in all branches of science, including engineering, chemistry, math, medicine, and quantum theory.50 The African Americans Lewis Howard Latimer, Elijah McCoy, and George Washington Carver, for example, w ­ ere born around the mid-­ nineteenth c­ entury and are now in the National Inventors Hall of Fame. Latimer was born f­ ree in 1848 Mas­sa­chu­setts to a run­away slave. He taught himself the art of drafting and drafted the designs for the telephone while working for Alexander Graham Bell. He l­ ater made an impor­tant improvement to the light bulb while working for a com­pany that rivaled Edison’s before getting hired by Edison himself.51 McCoy’s parents escaped slavery and settled in Canada, where Elijah was born. He lived briefly in Edinburgh, Scotland, as an apprentice to an engineer, before moving to Michigan, where he quickly applied inventions to the benefit of railways, before embarking on a successful ­career as a serial inventor.52 Carver was born into slavery to an idiosyncratic German American who supported the Union in the Civil War despite being a slave own­er.53 He raised George as his own son and provided the education that eventually led to his academic success as an agricultural scientist and inventor.54 ­These men ­were not alone. Surveys in 1900 and 1913 by the USPTO of patent attorneys, newspapers, and manufacturing firms identified over one thousand black inventors who held patents, including black w ­ omen such as Miriam Benjamin and Sarah Boone who w ­ ere born into slavery but educated in the Northeast.55 Many more could have registered for patents, according to the research conducted by Henry Baker of the Patent Office, if not for lack of funds to pay attorney fees.56 Another set of black inventors chose to have the white attorney hold the patent in his name so as not to depress the value of the invention as a result of its association with a black person. The African

Group Differences in Skills  •  165

American inventor of a clothes drying machine, Ellen Eglin, was explicit about this in an 1890 interview.57 By matching patents to census rec­ords, a team of University of Chicago economists estimated that 1.8 ­percent of U.S. inventions during the Golden Age of U.S. Invention (1880–1940) ­were made by black p­ eople, while another group of economic historians put the figure closer to 6 ­percent (for the 1870–1940 period), showing variation across de­cades from 3 ­percent to 10 ­percent.58 Overall, t­ hese figures suggest that black p­ eople ­were underrepresented in patenting ­because they comprised 10 ­percent of the U.S. population in 1920. Yet, black educational and occupational opportunities ­were severely restricted in the South, home to 82 ­percent of the black population. Outside of the South, black p­ eople comprised just 2.7 ­percent of the U.S. population, so the estimates above suggest that northern black Americans ­were about as likely to patent as white Americans—­perhaps even considerably more likely.59 Economic historian Lisa Cook finds that most black inventors lived outside the South and, if born in the South, migrated elsewhere.60 Moreover, the 2 ­percent to 6 ­percent estimates are biased downward, ­because as mentioned above, black ­people often had to file patents ­under the names of white ­lawyers. The unmistakable fact is that during one of the most impor­tant eras of invention in world history, African Americans played an impor­tant role comparable to the average white American. This is all the more incredible given that most of the U.S. black population had been enslaved for many generations. Even including Southern black p­ eople, the number of patents per person was higher among black ­people than Asians or any other racial group except white Americans during this period, according to the Chicago economists.61 It is quite pos­si­ble that black p­ eople outside the South patented at higher rates than Jews, but current data cannot test this directly. The patents rec­ords also show that black inventors w ­ ere more productive over their c­ areers—­patented more frequently—­than non–­black p­ eople and that patents held by black ­people w ­ ere cited more frequently by other patents, which is commonly regarded ­today as an indicator of higher quality.62 Even slaves contributed to innovation, though neither slaves nor their ­owners ­were allowed to hold patents for the inventions of slaves. None other than Jefferson Davis, the president of the Confederacy, filed for a patent on the basis of an invention for a boat propeller that he attributed to his slave,

166 • Chapter 6

Benjamin T. Montgomery. Davis l­ater promoted legislation in the Confederate Congress to pass a law allowing slave ­owners to own and patent the inventions of their slaves.63 Another example comes from Solomon Northrup, who was born ­free in upstate New York and was ­later abducted and sold into slavery. His autobiography was eventually adapted into the Acad­emy Award–­winning screenplay Twelve Years a Slave, but one detail is pertinent ­here: Northrup saved his owner a large sum of money by applying his engineering skills to devise a cheaper river-­based means of transporting the raw timber that was imported into the plantation.64 Another well-­ documented and far-­reaching example comes from the slave Onesimus, who introduced the practice of inoculation against smallpox to the United States in 1721 by explaining the West African practice to his owner, Cotton Mather, a prominent physician living in Boston.65 This knowledge saved the lives of countless Americans. It would take longer for African Americans to gain a foothold in academia, but Jews and Asians ­were also rarely professors or professional scientists in the early years of the twentieth ­century. An analy­sis of the 1,000 top U.S.-­ based scientists at the beginning of the twentieth c­ entury found that just 12.6 ­percent w ­ ere foreign born, and almost none were born in Rus­sia or China (0.6 ­percent and 0.2 ­percent, respectively).66 Most of the foreign-­born ­were from E ­ ngland and Canada. U.S. states with high literacy rates and robust public education systems—­such as Mas­sa­chu­setts and Connecticut—­were the most highly likely to foster scientific leaders. Mokyr explains the l­ imited Jewish contribution to the Industrial Revolution as resulting from rigid adherence to tradition and authority, which sapped them of creativity as applied to science and engineering. Another explanation is that u­ ntil the twentieth c­ entury, very few Jews or Asians grew up u­ nder republican governments that w ­ ere committed to equality of opportunity, high-­quality secular education for the masses, and property rights that would have secured the fruits of invention. A strict adherence to ancient customs at the expense of innovation would have been understandable u­ nder such conditions. Britain and especially the United States, by contrast, created the conditions for common ­people to venture and thrive. Eventually, u­ nder republican governments, Asian and Jewish Americans did thrive and went on to make countless contributions to American science and culture. ­Free black p­ eople w ­ ere oppressed in many ways, but for a brief time just ­after the Civil War and Reconstruction and before the full onset of Jim Crow

Group Differences in Skills  •  167

greatly curtailed the rights of black p­ eople, the pendulum of liberty swung in f­ avor of po­liti­cal equality. Following amendments to the U.S. Constitution that ended slavery, promised equality before the law, and granted black ­people the right to vote, Frederick Douglass was moved to say in 1870: We are a ­great nation—­not we colored ­people particularly, but all of us. We are all together now. We are fellow-­citizens of a common country. What a country—­fortunate in its institutions, in its fifteenth amendment, in its f­ uture. We are made up of a variety of nations—­ Chinese, Jews, Africans, Eu­ro­pe­ans, and all sorts. Th ­ ese dif­fer­ent races give the Government a power­f ul arm to defend it. They ­will vie with each other in hardship and peril, and w ­ ill be united in defending it from all its enemies, w ­ hether from within or without.67 Even with the diminished incentives afforded by racial discrimination, many black ­people saw this opportunity and took it. Madam C. J. Walker was born to a black ­family newly freed by the Civil War. Raised ­under Reconstruction, she went on to become an extremely wealthy entrepreneur.68 ­These achievements of African Americans should be kept in mind when considering the section that follows. As we ­will see, the best available evidence suggests that Northern black ­people had higher IQs than recent Jewish immigrants in the 1920s, which again, is impossible to reconcile with hereditarian theory. Efforts to Quantify Jewish Intelligence in the 1920s and Beyond

Aside from pointing out disproportionate achievement, hereditarian scholars claim that Jews have higher IQs. Compiling data from a few dozen small-­ scale studies, Lynn reports that American Jews’ IQ scores started at roughly the U.S. average for white p­ eople in the 1920s and 1930s (101.5), before gradually increasing to 107 during the 1940s, 1950s, and 1960s, and up to 111 ­after the 1970s through the early 2000s.69 The fact that Jewish IQ increased 10 points over 40 years in his database should be enough to eliminate the possibility that Jewish IQ advantages are ge­ne­tic in nature. Yet, he argues that it took time for the full nature of Jewish natu­ral genius to reveal itself: “Jewish IQs have increased significantly over this period. It appears that in the e­ arlier studies Jews w ­ ere handicapped, prob­ ably largely b­ ecause many of them ­were relatively recent immigrants who

168 • Chapter 6

­ ere impoverished, had poor nutrition and health, and many spoke Yiddish w as their first language.”70 Low social status and unfamiliarity with En­glish ­were real disadvantages for Jews, but Lynn does not acknowledge any similar environmental “handicaps” for black p­ eople suffering from Jim Crow, isolated and dispossessed Native Americans, former gentile serfs from Eastern Eu­rope, oppressed peasants from Mexico and southern Italy, or working-­class white ­people from the U.S. South. Nor does he explain why Yiddish speakers ­were so impoverished even as he and other (usually gentile) hereditarians extol the centuries of supposed wealth and high status of the Ashkenazi.71 Even in his ­limited compilation, Lynn makes crucial flaws in interpreting the studies he does include, as well as ignoring the prob­lem of se­lection bias. Most significantly, he ignores or dismisses the highest-­quality studies, all three of which contradict the claim that Jews had higher IQs before the 1960s, as I discuss in turn. Instead, he draws heavi­ly on small samples, where, in some cases, the psychologists conducting the study explic­itly give an interpretation to the data that does not match Lynn’s by pointing out the nonrepresentative nature of ­either the Jewish population or non-­Jewish reference population. Correcting for ­these errors, the studies Lynn cites provide no evidence that Jewish IQ scores ­were higher than those of native white ­people. IQ Testing in the Early Twentieth ­Century Contradicts Hereditarian Theory

More significantly, Lynn also dismisses the most famous study from the 1920s and one of the most comprehensive databases ever assembled on American cognitive testing.72 Understanding its implications for their theories, hereditarian scholars Rushton, Lynn, and Murray have all blithely dismissed the evidence from this research.73 The study was one of the most comprehensive IQ exams ever administered, and it was sponsored by the National Academies of Science. In 1919, a Harvard professor of psy­chol­ogy named Robert Yerkes led an effort to perform IQ exams on some 1.75 million American men—­those drafted into the military. Unlike most IQ exams before or since, this created data on a representative sample of the U.S. population and its subcomponents ­because the sample size was so large and the se­lection pro­cess was close to random, relying, as it did, on a military draft. At the request of Yerkes, Carl Brigham, a young scholar, led an analy­sis of a large sample of the results, which ultimately con-

Group Differences in Skills  •  169

tained data for 15,543 white officers, 81,465 native-­born white ­people, 23,595 black ­people, and 12,492 foreign-­born soldiers, of whom 2,340 ­were Rus­sian-born and at least half w ­ ere Jewish, according to Brigham. The IQ of Rus­sian immigrants came in at 86, almost a standard deviation below the IQ of native-­born white p­ eople, immigrants from Germany (100), and the Netherlands (103). Turks and Greeks outscored Rus­sians. Rus­sians scored only slightly higher than the average recently enslaved African American. For hereditarians, it gets worse. Northern black p­ eople, who w ­ ere slightly less persecuted by Jim Crow than t­ hose in the South, outperformed Jews by a comfortable margin (4 IQ points). Northern black ­people tested only slightly b­ ehind Irish, Austrian, and Turkish Americans and ahead of Greek, Rus­sian, Italian, and Polish Americans (­table 6.2). Northern black p­ eople, moreover, scored almost a full standard deviation higher than Southern black ­people (12 points), whereas Northern white p­ eople scored only 6 points higher than Southern white p­ eople. Thus, it would seem that moving north was of greater benefit to black p­ eople than white p­ eople in terms of enhancing educational opportunities. Other evidence from the period shows that black c­ hildren born in the South gained roughly 4 IQ points ­after moving north.74 None of the hereditarian critics denies the relevance of ­these data in the wider IQ debate, but they claim that Rus­sian Christians are mixed up with Rus­sian Jews and so t­ here is nothing to be learned about Jewish intelligence from the Rus­sian results. But this is badly wrong. First of all, Brigham is correct that the 1920 Census shows that 54 ­percent of ­people who listed Rus­sia or an imperial territory as their place of birth listed Yiddish, the language of the Ashkenazi, or Hebrew as their m ­ other tongue, but even this understates the ­actual percentage who w ­ ere Jewish. In 1936, according to data from the United States Census of Religious Bodies, t­ here ­were 4.6 million practicing Jews in the United States but only 54,000 members of the Rus­sian Orthodox Church.75 Thus, as of 1930, Rus­sian Christians would have constituted a tiny fraction of the 1.1 million American residents born in Rus­sia.76 More concretely, in 1885, the imperial Rus­sian government—­ which reigned over Jews in the Pale of Settlement—­decreed that only Poles and Jews ­were allowed to emigrate. The restriction seems to have been strictly enforced at least u­ ntil 1906. Up to that point, 93 ­percent of Rus­sian immigrants to the United States ­were Jewish, according to documents from U.S.

170 • Chapter 6

­ able 6.2.  Average intelligence test scores T for World War I recruits in 1918, by race for U.S.-born and by country of origin for foreign-born ­ ngland E Scotland Holland Northern U.S. whites Germany U.S. whites Denmark Canada Sweden Norway Belgium Southern U.S. whites Ireland Austria Turkey Northern U.S. blacks Greece Rus­sia Italy Poland U.S. blacks Southern U.S. blacks

105.8 103.0 102.9 101.7 100.6 100.0 99.6 99.4 97.5 95.9 94.9 94.4 92.4 92.1 90.8 90.7 90.2 87.3 85.5 84.1 82.4 79.2

Source: Carl C. Brigham. A Study of American Intelligence. (Prince­ton: Prince­ton University Press, 1923); For comparison, U.S. officers scores 128.4. Scores by region are from: Alper, T. G., & Boring, E. G. “Intelligence Test Scores of Northern and Southern White and Negro Recruits in 1918.” Journal of Abnormal and Social Psy­chol­ogy 39.4 (1944): 471–474.

port-of-­entry rec­ords analyzed by demographer Erich Rosenthal.77 From 1889 to 1914, the Jewish share of Rus­sian immigrants was at least 84 ­percent, and during this period, the majority of Jewish immigrants (72 ­percent) came from Rus­sia, with the Austro-­Hungarian Empire providing the largest secondary source. But w ­ hether the Rus­sian sample was 50 ­percent Jewish or, more likely, over 80 ­percent Jewish in no way means the data should be discarded or overlooked. ­These data are of much higher quality than all the studies described by Lynn (­because of their large size and representative sampling), and ­simple logic

Group Differences in Skills  •  171

shows that the maximum pos­si­ble group average IQ score for Jews was still quite low. Brigham reports detailed distributions of scores, and it is obvious that t­ here ­were very few outliers among the Rus­sian test-­takers. Only 3 ­percent of Rus­sians scored in the “A” and “B” group, which reflected the highest scores, compared to 12 ­percent of all test-­takers. If only Rus­sian Jews scored in the top group, and they ­were 50 ­percent of the total, then 6 ­percent of Rus­sian Jews scored in the top groups, still half the rate of all other test-­takers.78 Another issue is ­whether or not ­these IQ tests—­which ­were developed by the nation’s leading IQ scholars—­were biased against Jews more so than other groups. As Brigham argues, the testers went to ­great lengths to assign p­ eople tests based on their language proficiency and literacy. Th ­ ose who w ­ ere illiterate in En­glish w ­ ere assigned the “beta” version of the test, which relied entirely on pictograms and other wordless exercises that tested cognitive ability. Moreover, t­hose who did poorly on the English-­based “alpha” version—­which was based on the still-­used Stanford-­Binet IQ test—­were often given another chance with the beta version, just in case they ­were misidentified. Rus­sian Jews did very poorly in both alpha and beta, even compared to other immigrant groups from non-­English-­speaking countries. No ­matter how awkward for hereditarians, the conclusion that Jews born in the former Rus­sian Empire had very low IQ scores in the 1920s is inescapable. Another massive study ignored by Lynn and hereditarians was conducted by Charles Berry in the early 1920s. Working with the Detroit Public School District, he collected IQ data on 10,000 first-grade students, most of whom ­were age six. He reported the data by country of ­father’s birth and by language spoken at home. While IQ scores cannot be calculated based on his summary data, his report of the percentage of students who scored at dif­fer­ent levels shows that only 13 ­percent of Yiddish-­speaking ­children scored in the top 20th percentile on the IQ tests, whereas 23 ­percent of black ­children and 30 ­percent of white ­children who spoke En­glish at home scored in this top group (and 29 ­percent of all U.S.-­born white c­ hildren).79 A higher percentage of Yiddish speakers ­were in the top quintile compared to Italian speakers (6.4 ­percent), Polish speakers (10 ­percent), and Rus­sian speakers (10.7 ­percent), but not German speakers (16.7 ­percent) or t­ hose from miscellaneous language groups (15.6 ­percent). Again, this is unambiguous evidence that Jewish IQ scores ­were well below ­those of native white ­people, and consistent with Brigham’s data, Northern black p­ eople outperformed Jews in the same large

172 • Chapter 6

urban school district. In fact, black students outperformed e­ very group in which ­children spoke a foreign language at home, even though the test supposedly required no knowledge of written En­glish. Cognitive Testing in the Mid-­Twentieth ­Century

Sometime between the mid-­to late twentieth ­century, t­ here is evidence to suggest that Jewish Americans began to score consistently higher than black ­people and eventually native non-­Jewish white p­ eople on intelligence or other academic tests. The differences with white non-­Hispanic gentiles is accounted for by formal education. Young adults who identify as Jewish have the same IQ scores as white non-­Jewish young adults with the same education and parental education. This result is found in the only two large-­scale surveys that include both IQ and religion: the National Longitudinal Survey (NLS) 1979 (­people born from 1957 to 1964) and the NLS 1997 (born between 1980 and 1984).80 Returning to Lynn’s evidence from small, non-­random, and non-­ representative samples, he estimates that American Jews had a mean IQ of around 107 during the m ­ iddle part of the twentieth c­ entury, but the best available evidence shows that scores w ­ ere similar to native white peers 81 as late as 1960. Aside from study of the military tests, one other study of IQ stands out from the rest as large in scale and representative. It is called “Proj­ect Talent” and was led by University of Pittsburgh psychologist J. C. Flanagan, who founded the American Institutes of Research, which won a major award from the Department of Education to study U.S. students. In 1960, Flanagan and his colleagues collected IQ, demographic, and personality data on a random sample of 440,000 high school students in the United States (5 ­percent of the population).82 The test did not rec­ord religion but did ask ­whether the students’ parents could speak a number of non-­English languages. One category was Hebrew or Yiddish. As such, the question indirectly asked if the student’s parents w ­ ere Jewish. Certainly, one can imagine that not all Jewish c­ hildren born in 1945 had parents who still spoke Yiddish or Hebrew, but likely, it was the majority. A 1918 census report estimates that 85 ­percent of Jews living in New York City reported that Yiddish was their ­mother tongue.83 A 1922 study of ­children in the Detroit Public School District found that 80  ­percent of Russian-­born students spoke Yiddish at home.84 Fi­nally, in Melbourne,

Group Differences in Skills  •  173

Australia, a 1967 survey of Jewish ­children found that 76 ­percent said that at least one of their parents still spoke Yiddish, and 70 ­percent said that both spoke Yiddish.85 To distinguish families who speak a foreign language based on their ethnicity from ­people who may pick up a foreign language as a result of exposure, I focus the analy­sis on ­children who said their parents spoke the language fairly well, fluently, or very fluently. Nearly 22,000 c­ hildren of Yiddish speakers fell into this category, or 5 ­percent of the U.S. population of high school students, which is roughly the Jewish share of U.S. population at the time.86 Given the large sample size and random sampling method, this is arguably the highest-quality study of IQ ever conducted in the world for a single country. The results provide rather definitive evidence against the idea that ­either American Jews or Asians born around 1945 possessed above average intelligence. African Americans born into Jim Crow and raised in the West or Northeast registered IQ scores that w ­ ere only slightly lower than t­ hose from c­ hildren whose parents spoke Yiddish or Hebrew and higher than c­ hildren whose parents spoke “Oriental” languages. In fact, the scores of Northern black ­people ­were almost identical to ­those of Scandinavian immigrants. The c­ hildren of Yiddish speakers scored between .33 and .66 standard deviations lower than non-­Yiddish-­speaking white ­people. C ­ hildren selecting the racial category “Oriental” performed roughly the same as white ­people on the composite IQ tests, but t­ hose who said their parents spoke an “Oriental” language at least fairly well—­a much larger group than t­ hose who selected the race—­did quite a bit worse than white ­people (­table 6.3). The low per­for­mance of the ­children of Yiddish-­and Oriental-­speaking parents is not explained by the reliance on reading tests for nonnative speakers. Even on abstract tests such as visualization in two or three dimensions, both groups performed much worse than native white p­ eople. The designers of the Proj­ect Talent database suggested that the Abstract Reasoning subtest was “especially appropriate” for ­those who had just come to the country and had ­limited En­glish. On all of t­ hese tests, Jews and Asians performed well below average and only slightly better than their composite IQ score. Moreover, even third-­generation c­ hildren whose U.S.-­born parents still spoke Yiddish performed well below average. The same is true controlling for paternal occupation. Yiddish-­speaking ­children disproportionately reported that their ­fathers worked in professional occupations relative to white non-­Jewish

­ able 6.3.  IQ composite score by group for large sample of U.S. high T school students in 1960

IQ

Number of students tested

IQ by how well parents speak “Hebrew or Yiddish” Parents speak Hebrew or Yiddish rather poorly Parents speak Hebrew or Yiddish not very well Parents speak Hebrew or Yiddish fairly well Parents speak Hebrew or Yiddish fluently Parents speak Hebrew or Yiddish very fluently Parents speak any Hebrew or Yiddish (poorly or better) Parents speak Hebrew or Yiddish fairly well or better Parents speak Hebrew or Yiddish fairly well or better, at least one parent born abroad Parents speak Hebrew or Yiddish fairly well or better, both parents born in USA

84.6 82.2 87.5 90.7 95.6 89.5 91.9 88.0

3,397 3,942 6,263 5,468 9,961 29,031 21,692 8,909

94.7

12,783

93.9 100.0

308,460 126,128

93.6 90.2 88.7 86.7 89.9 80.2 73.4 85.3 99.2 83.9 87.4

39,090 30,151 28,454 21,131 17,812 15,991 3,480 1,399 975 312 235

IQ for other racial and linguistic groups Parents do not speak Yiddish or Hebrew White and parents do not speak Yiddish or Hebrew fairly well or better Parents speak German fairly well or better Parents speak Spanish fairly well or better Parents speak Italian fairly well or better Parents speak a Scandinavian language fairly well or better Parents speak Russian fairly well or better Parents speak Oriental language fairly well or better Blacks in the South Blacks in the Northeast or West Race is Oriental Race is Mexican American Race is American Indian

Note: IQ calculated from “IQ composite.” To calculate these scores, I first eliminated nonresponses (coded as 999), then calculated mean and standard deviations for the white non-Yiddish-speaking population for each age group, since there were multiple high school ages in the sample (with mean of 15). Then I converted individual scores to age-specific Z-scores before calculating group IQ averages. The white non-Jewish mean is calculated without using sample weights. Applying those weights barely lowers the mean score to 99.9. The question about parental language asks, “How well does your parent speak Hebrew or Yiddish?” Other questions have the same format but use the other languages listed. Source: American Institutes for Research, “Project Talent, Base Year Data, 1960,” May 23, 2015, Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2013-05-23, https://doi.org/10.3886/ICPSR33341.v2.

Group Differences in Skills  •  175

c­ hildren (12.7 ­percent vs. 6.3 ­percent), but while the non-­Jewish c­ hildren of professionals scored 108, the Jewish ­children of professionals scored just 101. Still, t­ hese results likely understate the IQ of all Jews b­ ecause more-assimilated Jewish parents—­those who no longer spoke Yiddish—­would likely hold higher social status and confer educational advantages to their c­ hildren.87 One study from Boris Levinson analyzed IQ data from 117 Jewish c­ hildren whose parents w ­ ere applying for a competitive Jewish private school in New York City. Since they ­were endeavoring to enroll their ­children in an expensive and competitive private school, ­these parents cannot be thought of as a representative sample of the broader Jewish population, but Levinson found that multilingual Jewish c­ hildren scored 4 to 5 IQ points lower than monolingual Jewish ­children. It follows that the scores above for the most-assimilated American-­ born Jewish c­ hildren would be roughly 100, if one added 4 to 5 points to account for their Yiddish-­speaking parents.88 In other words, Jewish c­ hildren scored right at the national benchmark for native non-­Jewish white c­ hildren. The more famous Coleman Report was also issued around this time and reported test s­ core data for large numbers of p­ eople. The report did not distinguish between Jews and non-­Jewish white p­ eople, but it did report the test scores of Asians (“Orientals”) in 12th grade. The average score across five dimensions suggested an IQ score of 97.2. White p­ eople scored higher on each dimension, which included math and nonverbal reasoning.89 While t­ hese data may seem startling to ­people familiar with hereditarian theory and the IQ lit­er­a­ture, the basic argument—if not t­ hese specific data—­ has been made by the economist Thomas Sowell in his review of The Bell Curve and his book on ethnic differences.90 Jewish and Asian IQ gains seem to have occurred largely for p­ eople born during and ­after the 1950s. In this context, one must also consider another ­factor in the sudden rise in test scores of Asian Americans and possibly Jewish immigrants from e­ arlier periods. Asian American immigrants—­like most immigrant groups—­tend to be far more educated than the native population of their home countries. This gap emerges, at least in part, as a result of immigration laws that allocate visas to university students or workers with a college degree. For example, 84 ­percent of current U.S. residents born in India and 55 ­percent of U.S. residents born in China who arrived ­after the age of 21—­and so did not grow up in the United States—­have a tertiary education compared to only 4 ­percent in their native countries (­table 6.4). The gaps are smaller for some African countries and for Haiti, ­because ­people from ­these countries are more likely

176 • Chapter 6

­ able 6.4.  Tertiary educational attainment rate of U.S. adult T immigrants from selected countries compared to adult populations in countries of origin

Country of birth

College attainment rate of U.S. residents aged 30–50 who arrived at age 21 or older

College attainment rate in country of origin, aged 30–50

India Nigeria China Taiwan Iran Hong Kong France United Kingdom Ghana Brazil Japan South Korea Haiti Honduras Guatemala El Salvador Mexico

84% 59% 55% 80% 61% 69% 72% 65% 38% 40% 56% 64% 20% 8% 7% 7% 6%

4% 2% 4% 30% 15% 26% 30% 26% 4% 7% 30% 45% 5% 5% 9% 11% 15%

Note: Data sorted by difference between educational attainment of emigrant population and homecountry population. Source: Data for U.S. residents are from IPUMS USA, using the 2012–16 American Community Survey. Data is restricted to people born outside the United States who arrived no earlier than age 21 and who are currently aged 30–50. Data from country of origin is from the Gallup World Poll and pools data from 2007–17 for the population aged 30–50.

to gain entry into the United States through refugee programs or family-­based programs that do not require high levels of education. For immigrants from Mexico and Central Amer­i­ca, the gap is reversed. Through a combination of nonsanctioned entry over the border and visas granted for less-­educated seasonal workers, Central American residents who enter the United States are far less likely to be college educated than t­ hose who remain in their native country. Thus, given the many pathways through which ­people migrate and their motivations for ­doing so, one must be cautious about comparing test scores and social status of immigrant groups and their native-­born c­ hildren and using that information to draw inferences about a global population.

Group Differences in Skills  •  177 GLOBAL TEST SCORE PER­F OR­M ANCE

Hereditarians like to point to global differences in literacy, test scores, and, where l­ imited data exist, formal IQ scores as evidence that Asians are naturally more gifted than o­ thers, but the data are more complicated than they have admitted and ultimately contradict the view that t­ here is any intellectually dominant race or ethnic group. The educational achievement scores of 15-­year-­olds enrolled in secondary school who take the OECD’s Programme for International Student Assessment (PISA) exam illustrate some of ­these issues. In general, Japa­nese and Korean ­children do very well, but roughly the same as Canadians, Estonians, Finns, and Americans living in Mas­sa­chu­setts. Likewise, Chinese perform almost as well as ­these other groups, but the data come from only four provinces: Beijing, Shanghai, Jiangsu, and Guangdong. Eigh­teen p­ ercent of the Chinese population lives in ­these four provinces, but 43 ­percent of this population has a post­graduate education. So, it is already an extremely elite group who lives in ­these places, compared to the rest of China.91 Data from the Gallup World Poll illustrate this. From 2012–2016, World Poll data show that 66 ­percent of the Chinese population aged 30 to 50 had only an elementary education. By contrast, in the 2015 OECD PISA database, only 18 ­percent of Chinese parents had an elementary education—or 23 ­percent if you apply their statistical weight. By contrast, in the United States, the Gallup data line up with the PISA data: only 6 ­percent of U.S. parents in the PISA database had only an elementary education, which nearly matches the Gallup World Poll result of 4 ­percent for 30 to 50 year olds in the United States. A further prob­lem with Chinese data is that one has to be very careful about analyzing data by region and applying it to the country as a w ­ hole. That is ­because Chinese cities—­including Hong Kong—­carefully manage migration from rural areas to cities, and public ser­vices such as education are restricted to ­those who have approved l­egal status. In Shanghai, this eliminates roughly half of the population from participation in public high schools, and many of the poorest c­ hildren return to their rural homeland as a result.92 The Chinese PISA scores, therefore, are not even close to being a representative sample of the national Chinese population.

178 • Chapter 6

Even Singapore’s data are flawed. Singapore is currently the world’s top performer on the PISA. Its scores would suggest an IQ of 113, compared to 109 for Estonia and Japan. Some point to the fact that many students supplement their education with private tutors in response to the country’s high-­pressure testing.93 However, one should also consider that as of 2015, 11 ­percent of Singapore’s primary school students failed to gain admission into its high schools, and so they w ­ ere not included in the OECD PISA exams.94 That removed the worst-­performing students from the sample. Data from the OECD’s other large-­scale cognitive exam—­one for adults— do not suffer from ­these flaws b­ ecause the exam does not require participants to have any special po­liti­cal status—­such as eligibility for public educational ser­vices. The Programme for the International Study of Adult Competencies (PIAAC) allows for high-­quality analy­sis of cognitive per­for­mance by country. The data w ­ ere collected from 2011 to 2015.95 While China declined to participate, 346 ­people born in China but living in a developed country w ­ ere included. Moreover, residents of Japan, South K ­ orea, Singapore, and Indonesia w ­ ere explic­itly included. I calculated means and standard deviations within five broad age groups and converted the results into the familiar IQ scores with a mean of 100. The results indicate that adults aged 16 to 65 born in China but living abroad (in one of ten countries that report the country of birth for participants) have a mean IQ of 87.1. Roughly 80 ­percent live in e­ ither South K ­ orea or Singapore. By comparison, African Americans scored higher on literacy and slightly lower on math than this group. Hispanic Americans scored about the same on math and slightly higher on literacy. It is likely that second-­generation Chinese immigrants make considerable pro­gress, but that makes a ge­ne­tic argument hard to sustain. Except for Japan and ­Korea, other adults born in other East and South Asian countries also score below the global average, including Singapore (98.3), Malaysia (96.7), Indonesia (94.4), and India (88.9). Israel was also included on the exam and scored below the global average. Since Israel is not exclusively a Jewish state, the mean score for Israel is not the best available score for its Jewish population. PIAAC data also report scores by the language test-­takers speak at home. Hebrew speakers scored right at the international average of 99.6, and their score is estimated with a small margin of error b­ ecause of the inclusion of Israel in the sample. Hebrew

Group Differences in Skills  •  179

speakers score just ahead of Polish speakers (98.3) but significantly b­ ehind Korean speakers (100.5) and En­glish speakers (101.4). Meanwhile, Yiddish speakers performed well below international norms, with an IQ of 93.6, somewhat below Rus­sian speakers (95.8) and Hungarian speakers (97.2), but their margin of error was large. Out of all language groups with at least 50 test-takers, Swedish speakers did the best (108.6), but differences with Japa­nese speakers (108.0) ­were not statistically significant. Scandinavian and Dutch and Flemish speakers did very well overall. Native Finnish, Dutch/Flemish, Norwegian, and Danish speakers scored between 106.5 and 105. Speakers of southern Eu­ro­pean and western Asian languages did less well—­including Spanish, Croatian, Albanian, Arabic, Persian, and Turkish. In this context, it is worth reiterating the point stated in the previous chapter that the IQ scores of immigrants are much higher when they arrive as young ­children and that this illustrates the strong causal effect of exposure to high-­quality education. For example, Sub-­Saharan Africans who arrived before age ten in New Zealand—­itself a high-­scoring country—­scored as high as the Japa­nese and Swedes (108.1). In general, the results in no way suggest that Israeli Jews or East Asians possess uniformly high IQs relative to the global average for residents of developed countries. That makes it all the more difficult to believe that Jews and East Asians living in Amer­i­ca ­today—­who once had low IQs themselves—­ come from some elite racial group. For that m ­ atter, Scandinavian Americans did not perform exceptionally well on IQ tests in the mid-­twentieth ­century e­ ither, and, like Jews and Asians, they contributed relatively ­little to the Industrial Revolution, and yet Scandinavian adults living in Scandinavia t­ oday are among the highest scorers globally. This again, undermines any notion that Nordic ­people are genet­i­ cally gifted. Rather, their high achievements, like t­ hose of the Japa­nese, as well as those of con­temporary American ­Jews, Native Americans, Korean Americans, and Chinese Americans, are the product of exposure to republican institutions and high-­quality learning opportunities from an early age in both their familial life and formal schooling. Like East Asians and Jews, Scandinavians have a long po­liti­cal tradition of valuing early childhood education and literacy, which in their case comes out of their Protestant ­emphasis on biblical study.96

180 • Chapter 6 SELECTIVE BREEDING DOES NOT EXPLAIN THE RISE IN IQ

Another lynchpin of Cochran et al.’s theory is that the most successful Jews (and by extension Asians) had more c­ hildren, which gradually raised the IQ of the population. Could this explain the extraordinary rise in Jewish American intelligence from the early to late twentieth c­ entury? The answer is no. To be clear, no one has proposed that a population’s IQ could be radically transformed by ge­ne­tics over just a handful of generations. Even the most aggressive assumptions about the heritability of IQ and the degree of selective breeding rule that out. It would take hundreds of years to generate a standard deviation change through selective breeding. Moreover, empirical evidence about the degree of selective breeding rules out the idea entirely. According to the Proj­ect Talent database, Jews with higher IQs had significantly fewer siblings than ­those with higher IQs during the 1940s and 1950s. Even worse for hereditarian theory, IQ had a stronger negative effect on reproduction among Jews than it did among other groups. The same is true for Asians. Th ­ ese facts are fundamentally incompatible with a ge­ne­tic explanation for the rise in Jewish and Asian American test scores during the twentieth ­century. Ironically, hereditarian theorists such as Murray and Herrnstein ­were very concerned by this pattern of “dysgenesis,” by which they and ­others meant the gradual decline in population IQ as a result of parents of lower intelligence out-­reproducing parents with higher IQs. They devote an entire chapter to it in The Bell Curve as a warning about the decline of Western civilization. They estimate that this pro­cess had resulted in a loss of IQ of roughly 0.8 points per generation throughout the twentieth c­ entury. In fact, they believe the effect is even worse ­because racial groups with lower IQs have more c­ hildren, and they believe racial differences are largely ge­ne­ tic. It is extremely awkward for this theory that a­ ctual IQ ­rose roughly 7.5 points per generation throughout the twentieth ­century in almost ­every country. This pattern of a generally rising IQ for the national population and specific subgroups, even as less-­educated parents have more c­ hildren, illustrates the power that education and nonge­ne­tic cultural f­ actors have in shaping cognitive ability. As Eu­ro­pean Jews and Asian immigrants integrated into American society, their low test scores converged with and ­later surpassed ­children from less-­educated backgrounds.

Group Differences in Skills  •  181 EFFECTS OF NONGE­N E­T IC AND NONETHNIC CULTURAL PRACTICES ON IQ

The social scientists Samuel Bowles and Herbert Gintis have defined culture as “preferences and beliefs that are acquired by means other than ge­ne­tic transmission” and argue that culture affects social success and ultimately the fitness of genes.97 The notion that culture affects be­hav­ior can, however, be easily misinterpreted as suggesting that long-­standing quasi-­permanent ethnic traditions—­ passed on through many generations—­are a primary driver of social outcomes. In fact, as Bowles and Gintis’s definition suggests, culture can come from anywhere—­including po­liti­cal institutions—­and changes therein. The economic historian Deirdre McCloskey has presented a compelling argument that the primary force ­behind the Industrial Revolution in ­England and northern Eu­rope was the development of a culture that emphasized the dignity and re­spect of ordinary p­ eople—­bourgeois life and business.98 This was a cultural force not primarily tied to ­family traditions—­indeed, the radical changes in occupational and educational mobility ­were radical departures from f­amily traditions, just as the scientific and entrepreneurial breakthroughs of the Industrial Revolution ­were radical departures from previous thinking. To clarify the point, consider the history of En­glish literacy, which grew rapidly during the Industrial Revolution, according to a recent account by historian Adam Fox. In 1500, just 10 ­percent of adult men and 1 ­percent of ­women could write their name, but by 1642, 30 ­percent of men and 10 ­percent of ­women could write, and by 1760, writing ability was demonstrated by 64 ­percent of men and 39 ­percent of ­women. Such rapid large-­scale changes ­didn’t come b­ ecause En­g lish parents—­who w ­ ere largely illiterate—­were teaching their ­children at home as part of a ­family tradition. Rather, as Fox puts it: “New educational provision was both the cause of the effect of the development of religious pluralism, the changing face of the En­glish economy and the more fluid nature of the social order.” Religious beliefs—­ especially that reading the Bible was edifying—­combined with practical needs and “bourgeois values”—­including re­spect for the common dignity of ­others—­motivated w ­ omen such as Alice Heywood of Lancashire, who was the wife of a weaver, to promote literacy in her community. In the words of her son, “It was her usual practice to help many poore ­children to learning by

182 • Chapter 6

buying them books, setting to schoole, and paying their master for teacher, whereby many a poore parent blessed god for help by their ­children reading.”99 Culture is power­ful, but it is fluid, dynamic, and responsive. It is also social and po­liti­cal. Culture cannot be understood as the atomistic and idiosyncratic practices of families, who learned every­thing from their parents and their parents’ parents. The values and practices that sweep through nations, communities, and families arise from social circumstances and publicly agreedupon norms. Thus, when social norms and ideologies change, one should expect changes in ­family practices. The po­liti­cal status of ethnic or racial minorities is one such social norm that has profoundly ­shaped the United States and its culture. A Cultural War over Education

In the United States, t­ here can be no doubt that cultural views on race have had profound effects on the opportunities for dif­fer­ent sections of society. The Civil War could be seen as an experiment that dramatically but only temporarily changed the dominant cultural view t­ oward black p­ eople as it changed their po­liti­cal status. Following the end of slavery, the U.S. government occupied the defeated Southern states. White opposition to the po­liti­cal equality of black ­people was explic­itly based on the theory of black inferiority. “The black man is what God and nature and circumstances have made him. That he is not fit to be invested with t­ hese impor­tant rights may be no fault of his. But the fact is patent to all that the negro is utterly unfitted to exercise the highest functions of the citizen.”100 According to W. E. B. DuBois, the Southern planters who dominated politics in the region stridently opposed taxation to support public education, even of white ­people. “Education they regarded as a luxury connected with wealth.”101 They w ­ ere particularly hostile t­ oward the education of black ­people. The occupation of the South by the Northern military temporarily stripped white ­people who supported slavery of voting rights, and Southern opposition to the po­liti­cal equality of black ­people was held in check.102 One consequence was that even during the war, the Northern government made im­mense changes to U.S. education policy. Southern politicians had voted down efforts to establish public universities, but ­after the South’s secession from the United States, Northern politicians passed the Morrill Act, which

Group Differences in Skills  •  183

established some of the world’s leading research universities, including Cornell University, the University of California, Berkeley, Pennsylvania State University, the University of Wisconsin, Purdue University, ­Virginia Tech, and Texas A&M. Importantly, a­ fter the war, the U.S. government initiated a massive program to increase the literacy of the formerly enslaved population, which also had the effect of creating, for the first time, a system of mass public education in the South. Many observers of the time noted both the alacrity and aptitude that black ­people demonstrated for education. DuBois’s history of Reconstruction quotes the influential Booker T. Washington, himself a freedman who went to ­g reat lengths to acquire an education, on the “intense desire which the ­people of my race showed for education”: “Few w ­ ere too young, and none too old, to make the attempt to learn. As fast as any kind of teachers could be secured, not only ­were day-­schools filled, but night-­schools as well. . . . ​ The ­great ambition of the older ­people was to try to learn to read the Bible before they died.”103 All around the country, white Northern officers—­who found a fervor for po­liti­cal equality born out of their opposition to their ­enemy’s creed—­formed schools for black soldiers, black folks raised and sacrificed their own l­ imited money for literacy, and the U.S. government’s Freedmen’s Bureau, as well as Northern philanthropies, all pitched in financially and sometimes personally to give black p­ eople opportunities long denied to them. The French minister of education made some observations in 1870 about the first schools for freed slaves in Washington, DC: “I was well qualified to judge for myself of the differences in intellectual aptitude of the two races. I must say that I have been unable to discover any. . . . ​A ll of the teachers . . . ​ that I have consulted on that point are of the same opinion.” He recounts other visitors who ­were also impressed by the “intellectual acuteness” of the black students, including one En­glishman, who said: “Never in any school in ­England, and I have visited many, have I found the pupils able to comprehend so readily the sense of their lessons: never have I heard pupils ask questions which showed a clearer comprehension of the subjects they ­were studying.” At Oberlin College in Ohio, meanwhile, the French minister reported that “14 young colored girls w ­ ere in the most advanced classes, and they appeared in no way inferior to their white companions.” Twenty ­percent of the student population was black, and they w ­ ere well integrated

184 • Chapter 6

and respected by students and faculty alike, according to reports, and regarded as equals.104 During Reconstruction, while Northern educational values predominated, freed Southern black ­people embraced the culture of learning. Literacy rates for young black adults in the South (aged 20 to 30) went from just 17 ­percent in 1870 to 34 ­percent in 1880 and 71 ­percent by 1910—­still far ­behind the Northern black rate of 96 ­percent in 1910 but much closer than before Reconstruction.105 Impor­tant research from economist Trevon Logan finds that literacy among black ­people tended to grow faster ­under jurisdictions controlled by black politicians, illustrating the social power of po­liti­cal equality.106 Eventually, with Lincoln killed and the Klu Klux Klan unchecked, white opposition to Reconstruction, which is typically dated from 1865 to 1877, wore out white Northerners’ temporary zeal for republicanism, and they allowed Southern white p­ eople to snuff out the spark that emancipation had ignited. Black p­ eople w ­ ere barred from voting and thus from being politicians. Black ­people ­were shut out of the professions, eviscerating any financial rewards from higher education. Still, as we saw above, what was done could not be undone. Despite im­mense obstacles, black folks made enormous contributions to the Golden Age of American innovation, and the literacy and intellectual achievements of freed slaves and the second generation of freed ­people are still with us. The fervor for equality and education that briefly united a large segment of white Northerners with freed slaves and other black p­ eople could be regarded as a cultural force. It was tied to real relations of power, engaged in a war of ideas and princi­ples. Culture is the proper term for ­family members passing on literacy to their c­ hildren and encouraging them to become educated, and it is the proper term for a mass po­liti­cal and social movement dedicated to that end. The Culture of Literacy

­ ese society-­wide cultural forces—­which ebb and flow—­should be kept in Th mind when considering how individual families pass on literacy, motivations for success, and education. First, we know from adoption studies that f­ amily circumstances m ­ atter greatly to educational attainment. A fairly consistent result from adoption studies is that the characteristics of nonbiological parents have very large and

Group Differences in Skills  •  185

­ able 6.5.  IQ growth and adoption into a professional f­ amily for T severely dis­advantaged ­children

Occupation of ­adoptive ­father Occupation requires minimal skill or training Skilled blue collar or low-­level man­ag­er Professional or high-­level man­ag­er

IQ of ­adopted ­children, ages 4–6

IQ of ­adopted ­children, ages 11–18

IQ gains since adoption

77.8

85.5

7.7

76.5

92.2

15.7

78.5

98

19.5

Source: Duyme, Michel, Annick-­Camille Dumaret, and Stanislaw Tomkiewicz. “How Can We Boost IQs of ‘Dull c­ hildren’? A Late Adoption Study.” Proceedings of the National Acad­emy of Sciences 96.15 (1999): 8790–8794.

significant predictive power for outcomes like income and education.107 One especially clear study examined cases of abused or neglected c­ hildren who ­were tested to have very low IQs between ages four and six and w ­ ere then ­adopted into families of varying social class. The IQ of c­ hildren a­ dopted by ­fathers who worked as professionals or high-­level man­ag­ers shot up by 19.5 IQ points (1.3 standard deviations) (­table 6.5). Being ­adopted by a ­father with a low-­status job also raised IQ substantially (by 7.7 points), but not nearly as much. The implication is that neglected or abusive homes are disastrous for IQ, that normal working-­class homes represent a substantial improvement, and that the homes and access to schooling provided by professional families offer a very large cognitive advantage. ­There is further evidence that the social status and education of parents—­ rather than ethno-­specific culture—­explain differential development of IQ for Jewish and Asian Americans. The National Longitudinal Survey of Youth 1997 (NLSY97) collected data from a random sample of youth who w ­ ere 12 to 16 years old as of December 31, 1996. It included a mea­sure of cognitive achievement, the Armed Ser­vices Vocational Aptitude Battery (­ASVAB). ­People raised as Jews, Hindus, and Buddhists scored very highly on this test, which is also highly correlated with SAT test score per­for­mance. Yet, controlling for parental education, ­there was no statistically significant difference between Jews and t­hose with East Asian religions and all other groups. In other words, ­there is a ­simple explanation for the education and cognitive success of Jews and East Asian ­children: They have highly educated

186 • Chapter 6

parents. The ­children of Christian or other parents who are also highly educated attain essentially the same scores.108 This implies that ­there is nothing specific to the ethnic traditions of the parents that translates into high IQ. The question then becomes what do highly educated parents—­whatever their race or religious background—­have in common when it comes to parenting their ­children and preparing them to become highly literate and numerate young adults, and why do some less-­educated parents—­often immigrants or ­children of immigrants, but also Southern black ­people during Reconstruction, and northern Eu­ro­pe­ans during the Industrial Revolution—­motivate their ­children to invest so heavi­ly in education? A strong emphasis on early childhood education is, of course, an impor­ tant cultural ­factor that certain groups—­including highly educated parents—­ often pass on to their ­children. Recent survey data in the United States show that black and Hispanic parents of three-­year-­old ­children are less likely to incorporate nightly bedtime routines (11 to 10  percentage points) than white parents and much less likely to have a reading component to t­ hose routines (20 to 22 percentage points). Taken together, this means that black and Hispanic c­ hildren are read to much less frequently at home, on average (­table 6.6). This pattern plays an impor­tant role in the development of in­equality in IQ and school readiness. A large-­scale review of dozens of experimental studies concludes that efforts to boost vocabulary and reading ability have large effects on the IQ of young ­children.109 Likewise, ­there is strong and consistent evidence showing that IQ growth is faster when parents consistently read to their ­children.110 Being read to literally changes the brain. ­Children whose brains are scanned while being read to show heightened neurological activity in the regions of the brain associated with decoding meaning.111 The fact that some parents read more to their c­ hildren than o­ thers should not be thought of in terms of intrinsic ethnic practices. If that ­were the case, it would be very difficult to change parenting be­hav­ior, when in fact parenting be­hav­ior with re­spect to reading is extremely malleable. Something as ­simple as a regular text message (three per week over the school year) sent by a school reminding parents to read to their c­ hildren substantially increases both reading activity by parents and the test scores of their ­children (by roughly 5 IQ points).112 The effects ­were largest for black and Hispanic ­children, whose parents ­were less likely to be reading to them routinely before the experiment. ­These results ­were from a carefully controlled experiment.

Group Differences in Skills  •  187

­ able 6.6.  Nightly reading by parents to T three-year-olds by ethnic groups in the United States ­ ercent who regularly read or tell stories as part of P nightly bedtime routine Hispanic Black White, non-­Hispanic

36% 37% 60%

Source: Data are from 3,600 c­ hildren from the Fragile Families and Child Well Being Study, which oversampled families born to unwed m ­ others. Th ­ ese statistics are reported in Milan, S., S. Snow, and S. Belay. “The Context of Preschool Children’s Sleep: Racial/Ethnic Differences in Sleep Locations, Routines, and Concerns.” Journal of ­Family Psy­chol­ogy 21.1 (2007): 20–28.

Mailing books to the homes of low-­income families has also proven to be effective.113 For example, Dolly Parton, a famous southern white American country musician and actress, created a program called “Imagination Library,” which mails one book a month to families in eligible areas with c­ hildren ranging in age from birth to age five. One study concludes that participation in the program increased the percentage of kindergarten ­children who exceeded benchmark scores on a well-­established literacy test from 46  ­percent to 61 ­percent.114 Another study found that it increased IQ by roughly 6.5 points.115 Taken together, this evidence demonstrates that ­family culture is highly dynamic and readily responds to even modest social cues, prompts, and access to resources. Thus, segregation—­which cuts off middle-­class and low-­ income communities, including white ­people, from the cultural practices and often the resources of highly educated parents—­inhibits the natu­ral cultural transmission of successful practices across class and race and extends and reinforces the less helpful practices—­such as watching tele­vi­sion instead of reading or overvaluing the practice of sports relative to the practice of math. Since ­children internalize the values of their parents, early reading activities are also likely to promote studious be­hav­ior l­ater in life, when ethnic differences in effort are pronounced. Jewish high school students in 1981 reported spending 4.3 hours more per week on studying while away from school than non-­Jewish students and 3.4 hours less on watching tele­vi­sion, according to my analy­sis of the National Longitudinal Survey of Youth (­table 6.7).

188 • Chapter 6

­ able 6.7.  Hours spent studying away from school per week by T group in 1981 among U.S. high school students Jewish No parent has four years of college At least one parent has four years of college All

Immigrant

U.S.-born, non-­Jewish

13.1

7.6

7.2

12.3

18.7

10.1

12.5

11.3

8.1

Source: NLSY 1979. Restricted to currently enrolled students. Adds minutes and hours spent studying away from school last week and uses 1981 sample weight, which is the year the studying data was collected.

Avoiding tele­vi­sion can hardly be regarded as an ethnic trait passed on for centuries. Moreover, non-­Jewish c­ hildren of highly educated parents act much more like Jews in the way that they allocate their time, regardless of race or religion. The most extreme group is the ­children of college-­educated immigrants, who spend an extra 8.6 hours studying per week compared to U.S.-­ born gentiles.116 As it happens, studying predicts higher IQ scores and watching TV predicts lower IQ scores, so t­ hese studying habits may very well explain many of the per­for­mance differences between groups and give an indication of why the ­children of immigrants so quickly catch up and often surpass the native population in educational achievements. Th ­ ese findings are consistent with evidence that Asian Americans’ academic effort explains their higher academic achievements compared to those of white Americans.117 High levels of effort and enthusiasm for education have been documented among high-­ achieving African immigrants to E ­ ngland and the United States, though in both places their cultural history differs dramatically from that of East Asians.118 One explanation for why effort is higher in Asians, which may apply to Jews, Africans, and other groups, is that East Asian parents and c­ hildren are more likely than white p­ eople to believe that hard work—­not genes or natu­ ral ability—­explains success.119 One scholar who has investigated the question asks why Americans have shifted t­ oward crediting innate ability over effort: “It seems likely that the increased attention to ge­ne­tics and ge­ne­tic influences on be­hav­ior, disease, and ­human nature has contributed to the belief that ge­ne­tics m ­ atters most. New research on ge­ne­tic influences is reported quite frequently in the popu­lar press.”120 If so, research efforts to link

Group Differences in Skills  •  189

ge­ne­tics to social outcomes should be balanced across interdisciplinary approaches, as in the Social Science Ge­ne­tic Association Consortium, so that findings about ge­ne­tic effects are put into appropriate theoretical context when communicated to the media and public. ­There is clear experimental evidence from social scientists at Stanford and Norwegian universities that belief in one’s ability to learn—­a “growth mindset”—­affects effort and a­ ctual math per­for­mance, while a hereditarian or “fixed mindset” is poisonous to both effort and per­for­mance.121 In the experiment, Norwegian high school students w ­ ere classified as having a growth mindset or a fixed mindset for math ability, based on their agreement with statements such as “Your intelligence is something about you that you ­can’t change very much” or “Some ­people are good at math and other ­people ­aren’t.” Fixed mindsets w ­ ere more common in students with low grades and students who w ­ ere on a vocational track that would not lead to a bachelor’s degree. The students w ­ ere then randomly assigned to e­ ither a control group, which received material about brain function but not its malleability, or a treatment group, which received a scientific article about the malleability of the brain written for a general audience for this purpose. They w ­ ere asked to relate the article to their life and write about how they would use the information to advise a friend struggling in math. ­These are ­things that psychologists find help change mindsets. Then, a few weeks l­ ater, both groups w ­ ere asked to complete algebra exercises and reassess their mindset. ­People in the treatment group experienced a .56 standard deviation increase in their mindset (becoming more growth-­ oriented) and outperformed ­those in the control group (by about .29 standard deviations), even though both began the experiment with a fixed mindset. In other words, this is clear evidence that beliefs affect per­for­mance and that beliefs can be changed. A belief that IQ is largely ge­ne­tic is detrimental. The implication is that gaps in IQ between classes are reinforced and perpetuated by bad social science. This dynamic has likely played out among the En­glish working class. Recent qualitative studies have focused on the fact that ­children of the white working class are among the lowest-­scoring groups in the country and often suffer from low literacy, ­little parental engagement in education, and low self-­esteem.122 Interviews with parents, teachers, and students suggest that they believe they are destined to occupy the bottom rungs of society—­a view inconsistent with En­g lish working-­class culture in the nineteenth c­ entury and the culture of immigrant groups t­ oday who w ­ ere not

190 • Chapter 6

exposed to a twentieth-­century theory that devalued low-­status groups as naturally inferior. Perhaps, the American and En­g lish working classes—­whatever their race—­could be encouraged to adopt a growth mindset from the achievements of African immigrants such as the Imafidon f­ amily. A ­ fter moving from Nigeria to London, the Imafidons raised five c­ hildren who have all distinguished themselves with genius-­level academic per­for­mance. The oldest d­ aughter, Ann Marie Imafidon, broke the rec­ord for passing the A-­level computer science exam at age 11. She went on to gradu­ate with a master’s degree in computer science and mathe­matics from Oxford at age 20.123 Her younger siblings had similar accomplishments. Peter and Paula passed A-­level math at age seven, breaking the En­g lish rec­ord and accomplishing a feat attained by only 22 ­percent of students ten years old or older; they ­later passed University of Cambridge advanced math exams at age eight.124 Peter and Paula inspired and even tutored another black girl from their church named Dee Alli, who became the youn­gest person to ever pass the math General Certificate of Secondary Education at age five.125 Thus, social contact with the f­ amily seems to have raised the achievement of o­ thers in their community. ­These West African immigrants—­with IQ scores that would effectively be off the charts—­did not come from a cultural tradition with centuries of literacy or urbanization. The dominant “cultural” perspective that the ­father, Chris Imafidon, seems to emphasize is a complete rejection of the ge­ne­tic basis for intelligence: “All ­children are born bright” is his mantra, which he expresses in vari­ous other formulations on a website he has created to inspire learning.126 This philosophy almost certainly contributed to the success of his ­children, and it is available to all of us. It is also much closer to scientific real­ ity than hereditarianism. This is not to say that mere belief or a growth mindset is sufficient to generate upward mobility. Far from it. ­Children need real, expensive, tangible investments from their parents, community, and governments to learn to read and write or acquire any other advanced skill. When parents lack the resources to invest heavi­ly in their child’s education—­for want of time, education, or money—­then the community and government need to fill the gap. The next chapter discusses the role of neighborhoods in shaping childhood outcomes and how too often po­liti­cal in­equality eviscerates chances to obtain equal opportunity.

7 Unequal Access to Housing Markets

AFRICAN AMERICAN SEGREGATION

Like Jewish ghettos in sixteenth-­century Italy, African American ghettos in the nineteenth-­century United States w ­ ere created, developed, and sustained by government policies, rather than simply voluntary private action. Thus, one of the most impor­tant barriers to racial equality in the United States—­ residential segregation—­has been the direct result of po­liti­cal in­equality generally and lack of equal access to housing markets, specifically. It is also impor­tant to note that exclusionary policies are not merely past abuses but are active ­today. Even now, government action and inaction are contributing to racial in­equality by preventing African Americans from participating on an equal footing in housing markets. This argument does not completely deny the importance of decisions made by private individuals, which shape market outcomes ­under the influence of government policies. Markets, government action, and legislation all require participants, and if ­those participants value racial discrimination, then market be­hav­ior and laws ­will reflect that. Accordingly, on the side of markets, white p­ eople have shown reluctance to live near black p­ eople, at least in large numbers. In research I conducted with Andre Perry and David Harshbarger of The Brookings Institution, we estimated that con­temporary U.S. homes lose roughly one-­quarter of their market value by virtue of their location in a black neighborhood.1 Our analy­sis suggests that home buyers undervalue property in black neighborhoods. Real estate agents and private groups have also played a role in creating or sustaining segregation, as I discuss below. With that being said, private actions ­were not nearly as impor­tant in explaining segregation as public actions, and t­hose public actions w ­ ere often directly in conflict with market dynamics. Moreover, a government committed to equal-

192 • Chapter 7

ity and market access would penalize and curtail private threats to equal participation, which did not happen throughout most of the twentieth c­ entury. The historic data on segregation bears this out. A detailed study by urban economists on the history of segregation across U.S. metropolitan areas concludes that government collective action, not decentralized private racism, created black ghettos.2 This is also the conclusion of the leading historical accounts of the ­causes of segregation, with eminent scholars such as Douglas S. Massey providing one of the clearest and most forceful accounts, and recent investigations by the journalist and writer Ta-­Nehisi Coates and the scholar Richard Rothstein also bearing this out.3 Moreover, the erroneous theory that extreme segregation is inevitable ­because black ­people prefer to live in neighborhoods with dif­fer­ent racial makeups than white p­ eople has been debunked by history and detailed empirical analy­sis.4 Survey data, as well as ­actual be­hav­ior, are quite clear that black p­ eople prefer neighborhoods that are far more integrated than their ­actual living arrangements tend to be, and con­temporary white ­people do not flee from or refuse to live in neighborhoods simply b­ ecause a few black residents live t­ here.5 Black-­white segregation was moderate in the early twentieth ­century, even though racism against black p­ eople was terrifically high. Segregation peaked much l­ ater—in the 1970s—­even as overt racial discrimination waned and racist ideology lost traction. T ­ oday, despite some pro­gress t­ oward greater integration, black-­white segregation remains extremely high (figure 7.1).6 Even into the twenty-­first c­ entury, black p­ eople remain relatively unlikely to have white neighbors, and white ­people tend to live in overwhelmingly white neighborhoods. In 2010, when non-­Hispanic white ­people constituted 64  ­percent of the U.S. population, a substantial number of white ­people—35 ­percent—­lived in extremely isolated census tracts, in which white ­people comprised 90 ­percent or more of the population.7 Meanwhile, just over 30 ­percent of residents ­were white in the census tract of the average black or Hispanic American. Asians tended to be more integrated, with roughly half of their neighbors consisting of white p­ eople for the average Asian American. The increasing population share of Asians and Hispanics over the last 30 years has increased the probability that a randomly selected neighbor w ­ ill also be Asian or Hispanic. This dynamic likely explains why Hispanic exposure to white p­ eople—­measured as the share of their neighbors who are non-­Hispanic white ­people—­has declined relative to 1980 (figure 7.2). As of 2010, Hispanics and black ­people have converged in their exposure to white neighbors at a low level. Yet, Asians and Hispanics are spread as

Unequal Access to Housing Markets  •  193 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Black share of population living in neighborhood of average black resident (isolation index) Percentage of black residents who would need to switch neighborhoods with non-black residents to balance population shares (dissimilarity index) Figure 7.1. ​African American neighborhood segregation from 1890 to 2010 Sources: David M. Cutler, Edward L. Glaeser, and Jacob L. Vigdor, “The Rise and Decline of the American Ghetto,” Journal of Political Economy 107.3 (1999): 455–506; Steven Manson, Jonathan Schroeder, David Van Riper, and Steven Ruggles, IPUMS National Historical Geographic Information System: Version 12.0 [1970 Decennial Census]. Minneapolis: University of Minnesota, 2017. http://doi.org/10.18128/D050.V12.0; John Logan, https://s4.ad.brown.edu/projects/diversity/Data/Download1.htm. See John R. Logan, and Brian J. Stults, “The Persistence of Segregation in the Metropolis: New Findings from the 2010 Census,” Census brief prepared for Project US2010 (2011).

evenly across neighborhoods as they ­were in 1980 (which is mea­sured by the index of dissimilarity). By this mea­sure, Asians are only moderately segregated, in that 42 ­percent would need to switch neighborhoods with someone from another racial group to balance out their distribution; for Hispanics, the corresponding number is 51  ­percent, and for black ­people it remains high, at 62  ­percent, which is nonetheless down from peak years. In summary, black segregation remains high and still exceeds that of all other groups, though Hispanic isolation from white ­people has converged to black rates a­ fter three de­cades of mass migration. This racial segregation means that black ­children grow up in the most disadvantaged neighborhood circumstances, relative to other groups. The Internal Revenue Ser­vice publishes data on the average income of families in nearly e­ very zip code in the United States.8 Zip codes are like neighborhoods and are created by the postal ser­vice to facilitate planning for mail delivery within a geographic area. Using Gallup survey data for roughly

194 • Chapter 7 100 90

88

84

79

80 70 60 50

62

57

46

30 20

51

40

40

74

47

35

28

31

31

1980

1990

2000

33 32

2010

10 0

Black

Hispanic

Asian

White

Figure 7.2. ​Average share of neighbors who are white by group, 1980–2010 Source: Data are from John Logan, available ­here: https://­s4​.­ad​.­brown​.­edu​/­projects​/­diversity​/­Data​ /­Download1​.­htm; see Logan, John R., and Brian J. Stults. “The Per­sis­tence of Segregation in the Metropolis: New Findings from the 2010 Census.” Census brief prepared for Proj­ect US2010 (2011).

half a million Americans, I calculated w ­ hether individuals who have ­children living at home are in a low-­income or high-­income neighborhood by comparing zip-­code-­level income to the median income of the larger local area commuting zones (which are like metropolitan areas but include rural counties).9 The results show that black p­ eople are the most likely group to be raising ­children in low-­income neighborhoods, followed closely by Hispanics. ­These groups also have very few p­ eople living in high-­income neighborhoods (­table 7.1). Even the c­ hildren of low-­income white p­ eople—­families making less than $36,000—­are raised in more-prosperous neighborhoods than the ­children of moderately high-­income black p­ eople ($90,000–­$179,000). On the other hand, upper-­income individuals, Jews, ­people with at least a bachelor’s degree, and Asians are highly likely to be raising their ­children in high-­income neighborhoods. Before Governments Created Ghettos

In the early twentieth ­century, leading white intellectuals, politicians, business leaders, and professionals ­were nearly unan­i­mous in arguing that black ­people ­were intellectually inferior to white ­people, and working-­class white

Unequal Access to Housing Markets  •  195

­ able 7.1.  Percentage of population with ­children living at home T who reside in low-­income or affluent neighborhoods, by group characteristic

Upper-income ­house­hold, ≥$180,000 Jewish Bachelor’s degree or higher Asian Moderately high income ­house­hold, >$90K–­$179K White Non-­Abrahamic religion Christian ­Middle-income ­house­hold, $60K–$89K Muslim Some college, up to two-year degree Moderately low income, $36K–­$59K Other race and multiracial White low-income ­house­hold, $90K–­$179K Low-income ­house­hold, 10 ug/dl) in blood of U.S. c­ hildren aged 5 and u­ nder, by race and year Sources: Mahaffey, K. R., J. L. Annest, J. Roberts, and R. S. Murphy. “National Estimates of Blood Lead Levels: United States, 1976–1980: Association with Selected Demographic and Socioeconomic ­Factors.” New ­England Journal of Medicine 307.10 (1982): 573–579; Jones, R. L., Homa, D. M., Meyer, P. A., Brody, D. J., Caldwell, K. L., Pirkle, J. L., & Brown, M. J. (2009). “Trends in Blood Lead Levels and Blood Lead Testing among US C ­ hildren Aged 1 to 5 years, 1988–2004.” Pediatrics, 123(3), e376–­e385.

Remarkably, in less than ten years the threat was eliminated for all but a small percentage of the population. The primary reason for the decline seems to be the mandatory removal of lead in gasoline starting in 1975, by order of the Environmental Protection Agency.25 By 1985, less than 50 ­percent of produced gasoline had lead; by 1990, it was roughly 0 ­percent. Concurrently, in 1977, the Consumer Product Safety Commission imposed new regulations to remove all but trace ele­ments of lead in paint.26 Still, recent data show that race gaps in lead exposure remain large (figure 8.1). Black ­children are three to four times more likely to have dangerous levels of lead exposure in their blood, using the cutoff of 10 ug/dl (though no safe exposure levels have been found). The historic and con­temporary racial and economic disparities in risk can be directly linked to economic and racial segregation, as a number of scholars have shown. Low-­income white ­people and, to a greater extent black ­people, have been compelled to live in older homes that w ­ ere more likely to have lead paint, less likely to have been remediated, and more likely to be located near dangerous environmental ­hazards.27 By concentrating toxic ex-

Group In­equality and Injustice  •  229

posures and impeding their remediation, government red-­lining and zoning directly contributed to well-­documented health risks. Beyond exposure to ­actual environmental toxins, dif­fer­ent neighborhoods also determine exposure to stress, which has its own well-­known toxic properties.28 ­People living in segregated communities are much more likely to be exposed to vio­lence and other known stressors, but this could be significantly reduced, if not eliminated, if p­ eople w ­ ere given an opportunity to participate in housing markets as equals. Evidence for this comes from a public housing proj­ect that was forced upon the mostly white community of Mt. Laurel, New Jersey, by the New Jersey state Supreme Court. In 1971, the court ruled that the town’s zoning laws ­were racially motivated to exclude black ­people. The remedy involved the construction of affordable housing units to allow ­people from nearby low-­income and mostly black communities (such as Camden) to move into Mt. Laurel. In 2000, ­after de­cades of court battles, the development, built by a nonprofit association, was fi­nally complete and began accepting residents. In 2009 and 2010, sociologists Rebecca Casciano and Douglas Massey surveyed ­people who had been accepted into the development by the nonprofit and compared their experiences to t­hose who had been wait-­listed or rejected for vari­ous disqualifying reasons but ­were other­wise similar.29 ­Because ac­cep­ tance into the program was based on the order in which the application was received and other criteria unlikely to be correlated with social outcomes—­ the study is quasi-­experimental. Living in the housing development significantly lowered the probability of experiencing a wide range of potential stressors and dangers—­including exposure to the sale or use of illegal drugs. The residents who moved ­were also far less likely to be arrested, robbed, or incarcerated. Perhaps, as a result, they ­were also less likely to report symptoms of stress—­such as fearfulness, trou­ble relaxing, or trou­ble sleeping (figure 8.2). Large-­scale national surveys confirm that black residents are much more likely to express concerns about their safety and security than other groups and generally express greater dissatisfaction with their home and community. A Gallup survey of 170,000 U.S. residents over the course of 2016 found that black p­ eople are twice as likely as white ­people and over four times as likely as Asians to strongly disagree with the following statement: “You always feel safe and secure.” Black p­ eople w ­ ere also much more likely to strongly disagree with statements that their home is ideal or that they c­ an’t imagine a better community to live in.30 ­These subjective judgments about neighborhood quality

4.3 Gangs

12.1

13.8

51.7

19.0

46.6

Drug Selling of Use of drugs paraphernalia drugs

15.6

55.2

Public drinking

26.7

Physical violence

22.4

Figure 8.2. ​Quasi-­experimental effect of exposure to disorder or vio­lence within neighborhood

Experimental group who gained access to affordable housing in white community

Control group who applied for housing project but had not been admitted

Prostitutes

38.8

48.3

65.5

Source: Casciano, Rebecca, and Douglas S. Massey. “Neighborhood Disorder and Anxiety Symptoms: New Evidence from a Quasi-­E xperimental Study.” Health & Place 18.2 (2012): 180–190.

6.0 Gunshots

38.8

Note: Data are from 2009 survey of current and former residents of an affordable housing proj­ect (Ethel Lawrence Homes) in New Jersey, as well as applicants to the program who w ­ ere e­ ither put on a waiting list or rejected. Propensity score-matching was used to address concerns that differences between the experimental group (­those who w ­ ere admitted) and the control group (­those who w ­ ere rejected or wait-­listed) w ­ ere meaningful.

Homeless people

13.8

52.6

64.7

Group In­equality and Injustice  •  231 20.3%

12.2%

6.9%

13.4% 10.9% 4.7%

White

11.8% 9.2% 4.5%

Black

7.2% 6.3% 3.3%

Asian

Hispanic

Strongly disagree with: You can't imagine living in a better community than the one you live in today. Strongly disagree with: The house or apartment that you live in is ideal for you and your family. Strongly disagree with: You always feel safe and secure. Figure 8.3. ​Gallup survey on the subjective feelings about neighborhood quality by race and ethnicity of adult respondent, 2016 Source: Gallup-­Sharecare Well-­Being Index. Data w ­ ere collected from January 2016 to December 2016. The survey is a random-­digit dialing sample of approximately 170,000 U.S. adults with ­either a cell phone or landline phone; responses are weighted to reflect demographic characteristics of the United States population and the probability of responding.

correspond to objective mea­sures, suggesting that ­people have an intuitive grasp of w ­ hether or not they live in a healthy neighborhood environment. ­After merging Gallup data with data from the Harvard Opportunity Insights proj­ect (created by Raj Chetty and his collaborators), I found that subjective mea­sures of neighborhood quality strongly predict higher intergenerational mobility (figure 8.3).31 “Better” neighborhoods are more expensive, which makes economic segregation a real prob­lem for families with lower incomes. Average neighborhood income is a very strong predictor of subjective neighborhood quality—­ measured as satisfaction with one’s community, home, and safety. Thus, 7 ­percent of t­ hose in the lowest third of neighborhoods strongly disagreed with the statement that they “always feels safe and secure” compared to just 3 ­percent in the ­middle third and 2 ­percent in the top third. ­These results classify neighborhoods based on average incomes relative to the local or metropolitan median income, but similar—­even larger gaps—­are found when all neighborhoods are simply ranked by income without regard to their broader local or metropolitan area (figure 8.4).

12.4%

8.9%

5.6% 3.7%

Middle third of neighborhoods

Strongly disagree with: The house or apartment that you live in is ideal for you and your family.

9.0%

3.2% 2.0%

Top third of neighborhoods

Strongly disagree with: You always feel safe and secure.

6.7%

Source of attitudinal data: 2016 Gallup-Sharecare Well-Being Index. Data were collected from January 2016 to December 2016, taken from a random-digit dialing sample of 94,128 U.S. adults with children living at home and either a cell phone or landline phone. Responses are weighted to reflect demographic characteristics of the U.S. population and the probability of responding. Source of zip-code income data: U.S. Department of Treasury, Internal Revenue Service, “SOI Tax Stats—Individual Income Tax Statistics—ZIP Code Data,” https://www.irs.gov/statistics/soi-tax-stats​ -individual-income-tax-statistics-zip-code-data-soi (accessed June 19, 2019). Source of commuting zone data: Median income by county from 2011–2016 Five-Year American Community Survey, American Fact Finder, U.S. Census Bureau, https://factfinder.census.gov/faces/nav/jsf/pages/index.xhtml. Commuting zones were matched to counties using a crosswalk created by David Dorn and made available by the Equality of Opportunity Project.

Note: Neighborhood income categories were determined by dividing average family income at the zip-code level by median family income at the commuting zone level.

Figure 8.4. ​Strong dissatisfaction with neighborhood, home, and safety by level of neighborhood income

Bottom third of neighborhoods

Strongly disagree with: You can't imagine living in a better community than the one you live in today.

16.8%

Group In­equality and Injustice  •  233

Segregation and Crime Victimization

The subjective concerns about neighborhood quality and safety have a corresponding objective basis in that black p­ eople are more likely than other groups to be victims of violent crimes—­such as sexual assault and robbery—­ and far more likely to be victims of hom­i­cide.32 This danger is a consequence of the unequal power and long-­standing oppression that created and sustains segregation. Even worse, the danger is exacerbated by the often fraught and unjust relationship that exists between the police, the criminal justice system as a ­whole, and the black community. The result is that segregation puts black ­people in excessive danger of being harmed by criminals and being harassed or arrested by police for be­hav­iors that are overlooked in white communities. I know the danger of segregated black neighborhoods from personal experience. Not that I was ever threatened or both­ered, despite spending a fair amount of my adult life living near or visiting segregated black neighborhoods. The dangers are much more pronounced for young black males, as I know from both statistics and experience. A few years ago, I signed up to participate in the Big Brothers/Big ­Sisters program in Washington, DC, which matches adult mentors to poor ­children, whose parents would like their ­children to have greater exposure to positive role models.33 I was assigned to a 13-­year-­old boy named Isaiah Harris, who lived in an apartment building in a segregated black neighborhood in DC with his younger ­brother Xavier and single ­mother Geneva Harris, who provided security ser­vices to downtown office buildings. Isaiah and I played basketball and chess, went bowling, went to DC’s Smithsonian museums, saw movies, and had meals together. Despite attending a m ­ iddle school with zero white ­children and failing test scores, Isaiah distinguished himself as a strong math student.34 He was one of only two students in his eighth-grade class to take advanced math courses at a nearby high school. One warm night in the spring of 2011, when he was 15, Isaiah left his block to talk to a girl he was interested in. A man who harbored a grudge against ­people from Isaiah’s street—­apparently related to gang rivalry—­de­cided he was ­going to kill someone—­any young man, it would seem—­from Isaiah’s street as revenge. Isaiah was murdered.35 The murderer had never met him, and Isaiah had never participated in any gang activity. This was a deeply traumatic experience for Isaiah’s b­ rother, his s­ ister who lived nearby, his many relatives, including his grand­mother who also lived in

234 • Chapter 8

the neighborhood, his classmates and teachers at school, ­people in his community, and especially his ­mother. At a time when American divisions are thought to be deep and unbridgeable and civic engagement non­ex­is­tent, I witnessed an outpouring of support for the ­family from ­people representing ­every race, class, and religion in the DC area. A candlelight vigil was or­ga­nized at Isaiah’s school, which gave the ­people who knew him best an opportunity to eulogize him and for the community to grieve collectively and reflect on the tragic loss. A few days l­ater, the funeral, which was performed at a neighborhood African American church, was widely attended and allowed for a more formal and ceremonial form of closure for the ­family and community. To help the ­family pay for the steep expenses of burial and the erection of a tombstone at a local cemetery, I assisted Geneva in setting up a nonprofit organ­ization with my local bank. A journalist was kind enough to publicize the address of the nonprofit publicly so ­people following local news could send in checks, and they did. My rec­ords show that 79 dif­fer­ent donors contributed a total of nearly $7,000 to the f­ amily; many w ­ ere my former colleagues at The Brookings Institution, as well as friends, some of whom had met Isaiah. Some ­were also connected to Xavier’s “Big ­Brother,” Jay, who also helped fund-­raise, and the Big ­Brothers organ­ization itself. Yet, the majority of the donors ­were strangers who had no relationship with the f­ amily. They often included kind, sympathetic notes. The DC government also provided crucial support that allowed Geneva to move out of the neighborhood to a safer nearby suburb. The fact that the killer was a black man in no way relieved the trauma and loss experienced by the Harris ­family and the wider community. Yet, justice was served, in that the killer was quickly arrested and given what seemed to be a fair sentence, and that process—­along with unan­i­mous support from the community—­did bring a sense of peace that has allowed the ­family to keep living, to establish and realize new goals, without ever forgetting the loss of their loved one. In too many other instances, however, black men have been killed not by violent gang members, but by the very ­people whose income is provided by the black community in order to protect them from such p­ eople: the police. ­Because of the recent proliferation of smartphones—­now equipped with video cameras and social media distribution networks—­a number of

Group In­equality and Injustice  •  235

controversial police encounters have been recorded and shared, providing evidence of injustice that other­wise would not have existed. In 2012, the bizarre death of Trayvon Martin became a focus of national debate. Trayvon, a young black man, was killed while merely walking home from the store, ­after being confronted by a man who had tried and failed to become a police officer—­George Zimmerman—­and fancied himself a neighborhood watch volunteer. The details are unclear, but apparently Zimmerman took it upon himself to drive up to Trayvon with his gun to confront what he thought was a suspicious-looking person. They fought and Zimmerman shot and killed him. Zimmerman was ­later acquitted of even accidental murder.36 Though Zimmerman was not a police officer, over a dozen cases over the next five years involved police officers who resorted to excessive force when confronting unarmed black boys or young men out of the very fear and suspicion that motivated George Zimmerman to perform what he thought was his civic duty by accosting a teenage boy g­ oing about his legitimate pedestrian business.37 Jim Crow is the f­ ather of ­these unjust encounters. The fear and prejudice that haunts police encounters with black men is in part the result of isolation. One of the strongest and most replicated findings in social science is called Contact Theory. It postulates that cooperation and trust between groups of p­ eople develop naturally from contact—­particularly when that contact is structured between fellow citizens with equal rights, rather than, for example, masters and servants or upper-­and lower-­caste members of society. In integrated communities, contact between groups arises from commercial interactions, informal exchanges, working relationships, shared extracurricular activities, or friendship. The evidence for its antiprejudicial benefits is overwhelming and cuts across time and country.38 In the context of the United States, ­people express more trusting attitudes ­toward ­others and are more likely to volunteer if they live in more racially integrated metropolitan areas.39 Segregation not only distorts the attitudes of police officers and white citizens, but also provides the practical opportunity to profile and control black ­people, just as it allowed for the control and oppression of Native Americans in the United States, Jews in premodern Eu­rope, and native black ­people in apartheid South Africa.

236 • Chapter 8

A recent analy­sis by economist Roland Fryer summarizes estimates from vari­ous databases and studies of police shootings. All agree that the probability of being shot by the police while unarmed is much higher for black ­people than white ­people. The relative risk ranges from 3.2 to 6.2—­meaning that young black men are three to six times more likely than young white men to be victims.40 In a misunderstood but impor­tant finding, Fryer finds that this gap is not explained by the fact that police act differently ­toward black p­ eople than white p­ eople a­ fter stopping them. Rather, the gap is explained by the much higher likelihood of police stopping black p­ eople. The over-­stopping of black men—­encouraged by policies called “stop-­and-­frisk,” which effectively assert that any young black man on the street should be ­stopped and searched—­seems to be the source of danger, and high-­density segregation makes the over-­stopping of black men more likely by concentrating them where they can be easily monitored.41 Other research from anthropologist Cody Ross of the Max Planck Institute shows that Fryer’s findings do not rule out the possibility that black suspects—­once encountered—­are treated inequitably with re­spect to use of force. In a mathematical model, Ross and his co-­authors show that a small number of strongly antiblack police officers could drive up the number of encounters with black ­people in situations that would never warrant lethal force, making it seem like the probability of using lethal force is the same for white p­ eople and black p­ eople, when in fact many of the encounters with black ­people never should have taken place.42 The Unequal Application of Criminal Enforcement

During the peak implementation years of the stop-­and-­frisk policy in New York City in 2011, local residents w ­ ere forcibly s­ topped and searched by the police 685,724 times. The overwhelming majority of t­ hese stops w ­ ere of black ­people (55 ­percent) and Latinos (32 ­percent). In 89 ­percent of encounters, the suspect was released with no charges.43 The searches ­were also concentrated in minority neighborhoods. In 2014, the top precinct for stop-­and-­frisk searches was the 106th in Queens. Only 4 ­percent of the population is white in this area.44 Defenders of “stop-­and-­frisk” and similar tactics say it lowers crime, even if it bothers innocent ­people. Th ­ ere is no obvious evidence that it lowers crime (crime trends show no change as stop-­and-­frisk waxed and waned), but even

Group In­equality and Injustice  •  237

if it did—­and, in theory, deter crime, as well—it would represent a gross miscarriage of justice through the unequal application of criminal enforcement and the fruitless harassment of innocent p­ eople.45 It is as if the goal is to punish 100 ­percent of crimes committed by black ­people but only 30 ­percent of crimes committed by white p­ eople. In fact, that disproportionate goal is not merely a hy­po­thet­i­cal example. The U.S. Department of Health and ­Human Ser­vices conducts a regular survey of health issues, drug use, and criminal be­hav­ior. It’s the best information available on criminal be­hav­ior that does not result in arrest. I focused on a sample of male respondents aged 18 to 30 ­because that age group encompasses the majority of arrests in almost ­every year since 1980, and since the late 1980s, arrest rates have exceeded 1 ­percent of the population for each age within the years 18 to 29, but are typically below that threshold for all other ages.46 Men are also far more likely to be arrested than w ­ omen. It turns out that black men in this age group are slightly less likely to sell or use drugs compared to white ­people, though the differences are too small to be statistically significant. Specifically, just 5.7 ­percent of black men aged 18 to 30 sold illegal drugs, compared to 6.1 ­percent of white p­ eople. P ­ eople who sell drugs frequently are also no more likely to be black. White ­people are slightly more likely than black p­ eople to have sold drugs ten or more times over the last year than black p­ eople (2.2 versus 1.8). Drug crimes among females are even more disproportionately white. Compared to white ­women aged 18 to 30, black ­women are less likely to use drugs (22 ­percent of black w ­ omen versus 28 ­percent of white w ­ omen) or sell them (1.0 ­percent versus 2.8 ­percent, respectively) drugs. Selling drugs, moreover, is hardly ­limited to the working class. White men enrolled in college full time are more likely to sell and use drugs than ­those who a­ ren’t enrolled. Fully 43 ­percent of white male college students reported using a serious illegal drug in the last year, and 6.9 ­percent reported selling them.47 Yet, black men aged 18 to 30 are 3.2 times more likely to be arrested for selling or using drugs than white men in the same age group. Combining arrest rates with selling rates suggests that 81 ­percent of black men who sell drugs are arrested compared to 24 ­percent of white men.48 Drug use by black ­people is also more aggressively policed. The share of young black men who report using drugs within the last year suggests 12 ­percent are arrested, compared to 4 ­percent of white ­people.

238 • Chapter 8

Using more sophisticated methods, I estimate that the odds of being arrested on drug charges for black males aged 18 to 30 is 3.8 times higher than for white males of the same age, controlling for specific age, lifetime drug use, drug use in the last year, and drug sales in the last year.49 For men over 30, black men are seven times more likely to be arrested on drug charges. This is what unequal protection ­under the law looks like. Th ­ ere is a systematic bias ­toward the punishment and prosecution of black crime relative to white crime. In New York City, 67 ­percent of ­people ­stopped for suspected marijuana possession and 75 ­percent of t­ hose suspected of selling marijuana ­were black, despite the fact that black ­people are just 25 ­percent of the New York City population.50 ­Those who think black ­people are more likely to lie about drugs than white ­people on anonymous government surveys ­will have to reconcile the facts of self-­reported drug use with the objective real­ity that non-­Hispanic white Americans died at higher rates from drug use and accidental poisoning from 2007 to 2016 than black ­people, according to the nation’s death rec­ords.51 Of course, ­people commit other types of crimes besides ­those involving drugs, and many of t­ hose other crimes have clear victims, who have e­ ither been physically harmed or had their property damaged or stolen. Black p­ eople are also more likely to be arrested for t­ hese crimes as well, and some people assume t­ here is a one-­to-­one correspondence between a crime and an arrest, but as we saw with drug data that assumption is grossly unfounded (figure 8.5). Compared to drug crimes, it is harder to mea­sure the population of ­people who get away with property or violent crimes, ­because surveys ­don’t typically ask ­people if they have recently murdered or raped anyone. The #MeToo movement has made it clear that sexual assaults are often unreported or other­wise do not result in arrest. In any case, murder and rape represent just 0.2 ­percent of all arrests in a given year.52 The National Survey on Drug Use and Health does not ask about murder or rape, but it does ask respondents to say w ­ hether or not they have stolen or tried to steal anything worth more than $50 in the last 12 months, and it also asks: “During the past 12 months, how many times have you attacked someone with the intent to seriously hurt them?”53 Relative to white ­people, black males aged 18 to 30 are less likely to say they have stolen or tried to steal something (2 ­percent versus 2.3 ­percent) and 2.4 times more likely to say they have attacked someone (4.9 ­percent versus 3.7 ­percent). Yet, black ­people are

Group In­equality and Injustice  •  239 0.4

37.8% 38.0%

0.35 0.3 0.25 0.2 0.15 0.1 0.05 0

5.7%

4.6%

6.1%

1.4% Drug arrest, last 12 months

Sale of illegal drugs in last 12 months Black

Use of common illegal drug in last 12 months

White

Figure 8.5. ​Rates of drug sale, use, and arrest of black and white men aged 18–30, 2013 Source: United States Department of Health and H ­ uman Ser­vices. Substance Abuse and ­Mental Health Ser­vices Administration. Center for Behavioral Health Statistics and Quality. National Survey on Drug Use and Health, 2013. ICPSR35509-­v3. Ann Arbor, MI: Inter-­university Consortium for Po­liti­cal and Social Research [distributor], 2015-11-23. https://­doi​.­org​/­10​.­3886​ /­ICPSR35509​.­v3. I defined common illegal drugs as marijuana, heroin, crack, cocaine, LSD, and meth. Drug sales question asked: “During the past 12 months, how many times have you sold illegal drugs?”

3.3 times more likely to be arrested for aggravated assault and 2.6 times more likely to be arrested for larceny-­theft.54 Again, the story is consistent with uneven enforcement of crime. Regardless of the apparent disproportionality with re­spect to nondrug crimes, the clearly unfair prosecution of black ­people for drug crimes has tremendous social consequences. For one, the sheer scale of drug arrests is astounding. The U.S. “War on Drugs” has been the predominant source of arrests and incarcerations in recent years.55 In 1980, only 7 ­percent of classifiable arrests w ­ ere for drug crimes, only slightly more than the share for violent crimes and far lower than the 21 ­percent share from property crimes. Since then, property crime arrests have plummeted and violent crime arrests have held steady, despite population growth and mass immigration. Yet, drug crime arrests increased from 0.6 million in 1980 to 1.6 million in 2014 and

240 • Chapter 8 25% 20% 15% 10%

0%

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

5%

Violent crime

Property crime

Drug crime

Figure 8.6. ​Share of U.S. arrests by type of crime for all classifiable offenses, 1980–2014 Source: Bureau of Justice Statistics, Arrest Data Analy­sis Tool.

3,000 2,500 2,000 1,500 1,000

Black

White

Figure 8.7. ​Arrests for drugs per 100,000 adults by race, 1980–2014 Source: Bureau of Justice Statistics, Arrest Data Analy­sis Tool.

2014

2012

2010

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

1982

0

1980

500

Group In­equality and Injustice  •  241

now account for 20 ­percent of all classifiable arrests. In ­every year since 1993, more ­people w ­ ere incarcerated for drug crimes than violent crimes. Nearly one-­third of total prison admissions since 1993 w ­ ere for drug crimes, the 56 largest source of admissions overall (figure 8.6). The arrests and incarcerations have damaged social mobility and economic opportunity for millions of Americans and their families. Th ­ ere is clear evidence from so­cio­log­i­cal research that the l­ abor market for ex-­offenders is difficult, and as my former colleagues Isabel Sawhill and Scott Winship have shown, getting convicted of a crime in adolescence greatly lowers the probability of reaching the ­middle class by m ­ iddle age.57 The eruption of drug enforcement activity has been particularly harmful for black ­people living in segregated communities. From 1980 to 2014, t­ here ­were 14 million drug-­related arrests of black p­ eople, averaging over 400,000 per year, and the disproportionate arrest rates for black ­people have ranged from 2.6 to 5.2 over the years (figure 8.7). Stepping back, we can see that neighborhood in­equality—­itself a monumental failure of democracies to protect market access—­has played a very impor­tant role in creating and sustaining an underclass of Americans. The tools of market manipulation and obstruction that I documented in the last chapter—­deed restrictions and then zoning—­have lowered the realized intellectual abilities and occupational opportunities of countless ­children by depriving them of quality schooling, isolating them from the cultural norms of the ­middle class, exposing them to environmental toxins and damage from the stress of social disorder, and punishing them excessively for crimes committed just as often by the upper-middle class.

9 Unequal Access to the Buying and Selling of Professional Ser­vices

This book started by linking the rise of far-­right nationalism to a strong sense of economic unfairness, which itself is linked to high and growing in­ equality, a distrust of liberal government, and an under­lying belief in racial superiority. I argued that the rise in in­equality and the slowdown in economic growth are closely linked to the increasing inefficiency in ser­vices and to the rising share of income ­going to elite professionals who provide ser­vices in healthcare, law, finance, and the like. I then sought to demonstrate that t­ here is no scientific basis for believing that racial or ethnic groups differ in any natu­ ral or fundamental way in their eco­nom­ically productive characteristics and that in a just society every­one should have an equal opportunity to succeed. This chapter discusses how elite professionals have managed to appropriate a significant proportion of national income for themselves. The core prob­lem is that professional associations have effectively taken over the governance of markets for many impor­tant products and ser­vices. ­These groups have a strong incentive to distort laws to the advantage of their members, and they have the influence to do so. The results are profoundly negative: Men and ­women who could ably and profitably provide goods and ser­vices are denied the right to do so. This situation violates a core principle of a just society—­that p­ eople should be able to pursue the occupation that best suits their nature, interests, and skills. Beyond that, the distribution of the good ­things in life—­such as health, income, and social status—is greatly distorted by the rule of professional elites. The rest of society pays dearly in order to shore up the incomes of t­ hese groups. This is not to say that elite professionals do not deserve to be paid well. As described in previous chapters, skills broadly understood—­including length of education, the natu­ral and developed traits of personality and cognitive

Unequal Access to Professional Ser­vices  •  243

ability—­reliably predict income to a greater or lesser degree across developed democracies. Professionals tend to be highly educated and demonstrate relatively high levels of cognitive skill. Th ­ ose with higher levels of education also score higher on the noncognitive skills that predict success in life (i.e., conscientiousness, enthusiasm, and emotional stability). Along ­these lines, as discussed in chapter 3, some portion of variation in earnings within democracies can be directly tied to “merit”—­the value an individual contributes to an organ­ization, a business, or a society—­and to the relative scarcity of the skills brought to bear in creating that value. Yet, as ­we’ve seen, most professions are consistently overpaid even ­after taking their skill levels into account. In many cases, this is the result of many years of po­liti­cal effort by the vari­ous associations that lobby for their interests. Over the last two de­cades, as developed economies have become more integrated, ­there have been several attempts to systematically compare the regulations of professional ser­vices across countries. In an OECD study published in 2000, the authors noted distinct ways in which professional groups hinder markets from operating efficiently and effectively, to the benefit of their professions and the expense of the public.1 Common control mechanisms used by professional groups across countries include (1) excessive entry barriers, (2) collusion in setting fees or deterring advertising that might lower prices, (3) limits on how firms are or­ga­nized and on the extent to which professionals can be employed by outsiders, and (4) the establishment of a l­ egal mono­poly on ser­vice provision. In many cases, t­ hese laws and practices are controlled by professional associations, which prac­ti­tion­ers are often required to join. Such associations also provide impor­tant information and networking opportunities to members, at the expense of outsiders. If that sounds similar to the way the craft and merchant guilds operated, it is no coincidence. The same antimarket tools have been used for centuries across many dif­fer­ent countries and economies.2

HEALTHCARE

Being a doctor pays very well in all rich countries. In Spain, Sweden, and Iceland, general practice doctors typically earn twice as much as the average worker. In the UK, Ireland, and Canada, it is three times as much.3 But doctors’ pay in the United States truly stands out: Physicians and surgeons earn

244 • Chapter 9

­ able 9.1.  Average pay of health professionals divided by average T worker pay, by country, 2000–2016

Belgium United States Austria Luxembourg Canada Australia Germany Netherlands Ireland United Kingdom Mexico New Zealand Chile Israel France Denmark Finland Slovenia Sweden Spain Czech Republic Iceland Estonia Hungary Latvia Poland

Specialists

General prac­ti­tion­ers

Nurses

6.0 5.0 4.6 4.5 4.5 4.2 4.0 3.9 3.6 3.5 3.2 3.0 3.0 3.0 2.9 2.6 2.5 2.4 2.4 2.3 2.1 2.1 2.1 1.8 1.5 1.5

2.1 3.6 2.9 2.9 2.9 1.8 3.8 2.3 2.8 2.9 2.6 2.5 2.4 1.9 3.0 2.9 1.8 2.2 2.2 2.1 3.1 2.1 1.8 1.6 1.1 2.1

1.1 1.3 1.4 1.1 1.2 1.2 1.1 1.0 1.1 1.7 1.2 1.7 1.3 1.0 1.1 1.0 1.0 1.3 1.0 0.9 1.0 0.9 0.8 1.0

Note: Numbers in table show relative pay, expressed as the ratio of average pay for the group shown to average pay of all workers in country. Source: Data from OECD.STAT. “Health Care Resources: Remuneration of Health Professionals,” accessed April 3, 2018. Salary ratios are averaged from 2000 to 2016 using all available data for each country.

4.9 times as much as the average worker.4 Only Belgium provides higher relative compensation to specialists, but the United States compensates general prac­ti­tion­ers much higher, so the average physician in the United States receives higher relative pay than a physician in any other OECD country (­table 9.1).5 Of 480 occupational categories in the United States, census data from 2012–2016 shows that physicians and surgeons possess, by far, the highest

Unequal Access to Professional Ser­vices  •  245

average income ($219,000), home values ($706,000), and probability of being in the top 1 ­percent of income earners (31 ­percent).6 ­There are several reasons why doctors in the United States are so overrepresented among top earners. U.S. physicians, more so than their Eu­ro­ pean peers, have created large entry barriers, including at the hospital level; they have more aggressively suppressed lower-­cost professional competitors; and they have eliminated more efficient ways of organ­izing health ser­vices. Let us start with entry barriers. High pay creates incentives for more ­people to enter a field. Yet, becoming a physician is an extremely lengthy and costly pro­cess. In the United States, it takes a minimum of 11 years of constant work and study to complete medical school, including a bachelor’s degree program, four years of gradu­ate study, and at least three years of residency training, depending on the specialization, plus state licensing exams. Typical medical school tuition, meanwhile, is over $200,000, which comes on top of the roughly $120,000 required to complete a bachelor’s degree.7 In the nineteenth ­century, it took only eight months of study to complete medical school and earn a doctorate in medicine.8 While eight months is surely too short a period, 11 years of formal study strikes me as excessive. ­These entry barriers lower the number of p­ eople willing or able to become physicians, thereby artificially raising the salaries of physicians. In 2016, the average salary for a ­family practitioner was $184,240, and that is among the lowest-­paid physician specialties. Anesthesiologists—­who provide pharmacological pain relief before and ­after surgery—­are the highest-­paid specialty, even ahead of surgeons, with average salaries of $270,000 (­table 9.2). Meanwhile, nurse anesthetists and nurse prac­ti­tion­ers are trained to provide essentially the same ser­vices as anesthesiologists and ­family prac­ti­tion­ ers, but their salaries are between 50 ­percent and 60 ­percent of those of their counter­parts. The nursing pathway requires four years of nursing school and licensure as a registered nurse, plus the completion of a master’s degree program in nursing, which has clinical and classroom components. Becoming a nurse practitioner or anesthetist does not require additional years of residency training for ­those who have experience working as a nurse. Thus, a nurse can start practicing five years ­earlier than a physician, without paying nearly as much in educational fees or foregone income from being out of the ­labor force. ­There is no question about the competency of nurse prac­ti­tion­ers at providing ­family and general practice medical care. ­A fter reviewing the extensive

246 • Chapter 9

­Table 9.2.  Salary comparisons of healthcare professionals Mean salary, 2016

Elite healthcare professions with high entry barriers Anesthesiologists Surgeons Obstetricians and gynecologists Physicians and surgeons, all other Internists, general ­Family and general prac­ti­tion­ers Psychiatrists Pediatricians, general Dentists

$269,600 $252,910 $234,310 $205,560 $201,840 $200,810 $200,220 $184,240 $178,670

Healthcare professions with lower entry barriers Nurse anesthetists Optometrists Nurse prac­ti­tion­ers Nurse midwives Chiropractors Dental hygienists Registered nurses

$164,030 $117,580 $104,610 $102,390 $81,210 $73,440 $72,180

Source: Bureau of L ­ abor Statistics, Occupational Employment Statistics, 2016. Source does not include self-­employed workers or ­owners of un­incorporated businesses.

medical lit­er­a­ture showing that patient outcomes are just as high for nurse practitioners as they are for physicians, the Institute of Medicine at the National Academies strongly recommended that states grant them full in­de­pen­dence to practice f­amily and general practice medicine.9 Likewise, a recent study by Dutch-­and British-­based healthcare experts reviewed 25 studies of nurse-­ practitioner efficacy, which used randomized controlled experiments—­the gold standard in social science—­and the conclusion was extremely clear: “The findings suggest that appropriately trained nurses can produce as high quality care as primary care doctors and achieve as good health outcomes for patients.”10 Nurse midwives are similar to nurse prac­ti­tion­ers in terms of their educational requirements and average pay, but focus on reproductive issues, birth, and early childcare. A review of the lit­er­a­ture finds that low-­to moderate-­risk home births attended by midwives are at least as safe as hospital births, ­whether attended by a physician or midwife.11 Indeed, births attended by midwives utilize far fewer costly interventions—­dramatically lower rates of

Unequal Access to Professional Ser­vices  •  247

epidurals, induction, and Caesarean sections—­with comparable mortality rates, and in this way midwife-­attended births would be far more efficient even if midwives ­were paid as much as their physician counter­parts.12 Patient safety, according to another study, is in no way harmed when nurse anesthetists work without the supervision of physician anesthesiologists.13 Despite this empirical evidence, many states impose heavy restrictions on the practice of nursing in ways that explic­itly benefit physicians. The District of Columbia and 22 of the 50 states allow nurse prac­ti­tion­ers to work in­de­ pen­dently of physicians in accord with their training and the interests of the public;14 27 states allow midwives to practice in­de­pen­dently; and only 16 states allow in­de­pen­dence from supervision for nurse anesthetists.15 Even worse, the richest and most populous states, such as California, Texas, New York, Florida, and Illinois, tend to be the most restrictive. The result is that the ser­vices provided by one elite professional group— to which access is extremely restricted—­are not allowed to be provided by less elite professional groups, even when ­those groups meet state licensing requirements and scientific evidence demonstrates their efficacy and competence. One can imagine that gender in­equality has more to do with this imbalance than professional mastery, given the historic barriers to w ­ omen becoming doctors but not nurses. The Physician Lobby

State po­liti­cal representatives are duty bound to represent the interests of their constituents, but they routinely side with the interests of elite professionals—­ such as doctors—­over the interests of citizens. It’s no mystery why. Representatives of physicians wield considerable power in opposing the interests of nurses, as well as the interests of the consumers and taxpayers who pay for healthcare. And t­ here’s no need to take my word for it. Physician lobbies are quite open about their anticonsumer advocacy and boast about their power to persuade legislators and the public that paying excessive healthcare costs is necessary to guarantee their health and safety. In 2012, a physician’s advocacy association, the Physicians Foundation, wrote a candid report entitled “Accept No Substitute: A Report on Scope of Practice.” As the title implies, the report identified potential competition from nonphysicians as a key concern for physicians. The report describes po­liti­cal momentum to grant greater scope of practice or autonomy to nonphysicians

248 • Chapter 9

such as nurse prac­ti­tion­ers, optometrists, and psychologists. It mentions pressure to lower costs as one of the primary motivations ­behind ­these changes as well as “the lack of hard evidence that physicians do in fact provide better care than non-­physician providers.”16 The report’s authors list concrete steps that their lobbyists are taking to protect the power of physicians in concert with the largest physician lobby of all, the American Medical Association (AMA): In the face of all ­these challenges—­both external and internal—­medical society leaders have taken a number of steps that include: strengthening their lobbying capacity, working more closely with specialty socie­ties, stepping up their fund­rais­ing and campaign contributions, fostering strong relationships with key legislators, and keeping the focus solely and relentlessly on patient safety. In addition, they have worked with the American Medical Association and other national medical organ­ izations through the Scope of Practice Partnership. The AMA, in turn, has been active on federal policy issues related to scope of practice; produced its Scope of Practice data series on ten non-­physician professions; launched a national Truth in Advertising campaign that has already generated new laws in fourteen states; and funded a major mapping study to document where non-­physician providers are actually practicing.17 The report makes it abundantly clear that the physician lobby, including the authors who are affiliated with the Physician Foundation, are stridently opposed to state laws that would lower the cost of healthcare and increase the efficiency of accessing it.18 Their motivation, to be fair, may very well be patient safety, as they understand it, but insofar as that is a genuine concern, I’d argue it is based on prejudice rather than science. What­ever the true motive, the report documents specific efforts by physician groups to uphold the following prohibitions: on optometrists treating glaucoma, on podiatrists treating ankles, on nurse anesthetists practicing in­de­pen­dently of physicians, and on pharmacists providing routine vaccinations. Citing anonymous interviews with medical executives, it quotes one as bragging about holding back the tide of consumer-­friendly deregulation by wielding “sheer po­liti­cal muscle.” Another boasted of a victory over nurse anesthetists: “We beat them back in the legislature.”19 The report notes with some satisfaction that only a tiny fraction of nurse prac­ti­tion­ers are operating

Unequal Access to Professional Ser­vices  •  249

in­de­pen­dently as a result of the po­liti­cal opposition of medical socie­ties. This same group laments the disappointing lack of evidence in its ­favor and suggests the need to commission studies that w ­ ill be more favorable to its mission. The California Medical Association—­the largest physician lobby in the state—­spent $13 million on state campaign finance contributions from 2010 to 2016. Total spending by physicians and their associations reached $26 million in the state over the same period. Nurses also have associations, but their po­liti­cal resources are dwarfed by t­ hose of physicians, even though ­there are far more nurses. Nurses and their associations spent only $1.3 million on campaign finance in California during the period.20 This physician spending was apparently effective. In 2015, California legislators introduced a bill that would allow nurse prac­ti­tion­ers to work in­de­pen­dently of physicians. On May 7, 2015, the bill passed the California State Senate by a margin of 25 to 9.21 All of the nine senators voting against the bill received substantial donations from the California Medical Association, which openly lobbied against it, saying it would “put Californians at risk.”22 Despite this initial success, on June 30, 2015, the bill officially died in the other legislative body, the California State Assembly, where it lost a vote in the Business and Professions Committee by a margin of nine to four. The average assembly member voting to move the bill forward received $13,000 from physicians’ organ­i­ zations, mostly the California Medical Association, whereas t­ hose voting to kill the bill received an average of $20,000 from t­ hose groups.23 The California Medical Association (CMA) was quite open about its hostility to the bill, calling it “dangerous” in a press release from the CMA’s president, Paul R. Phinney, M.D. In referring to the bill back in August 2013, when it passed a senate committee vote, Dr. Phinney stated: Patients should be deeply concerned that SB 491 was able to move forward ­today. If SB 491 becomes law, nurse prac­ti­tion­ers ­will be allowed to practice without any supervision by a physician, despite the enormous differences in education and training between the two. SB 491 allows nurse prac­ti­tion­ers to diagnose and treat patients beyond their capabilities and without any additional training, jeopardizing patient safety.24 Fi­nally, in 2017, the AMA’s House of Delegates formally passed a resolution with the intention of suppressing the right of nurses to practice medicine without physician supervision. It states that the AMA should

250 • Chapter 9

(1) effectively oppose the continual, nationwide efforts to grant in­de­ pen­dent practice to non-­physician prac­ti­tion­ers; (2) effectively educate the public, legislators, regulators, and healthcare administrators; and (3) effectively oppose state and national level legislative efforts aimed at inappropriate scope of practice expansion of non-­physician healthcare prac­ti­tion­ers.25 This is consistent with material on the AMA website in 2018.26 It might be considered malpractice if a physician knowingly prescribed a treatment that contradicts the scientific lit­er­a­ture, and yet, when it comes to public policy, t­ here is no accountability for a doctor—or association leader—­ publicly contradicting the scientific lit­er­a­ture by falsely stoking fear. Aside from nurse prac­ti­tion­ers, the AMA and other physician groups have also made enemies out of other allied health professions, including optometrists, who typically complete four years of gradu­ate school, ­after science-­based undergraduate training, which is similar to the premedicine track taken by physicians. Optometrists would like to practice in­de­pen­dently of physicians, prescribe medi­cation, and treat eye disease, but ophthalmologists, who go to medical school, argue that they are not qualified to so. As it happens, optometrists get paid roughly 60 cents for e­ very dollar earned by ophthalmologists, so the public would see huge savings if medical ser­vices shifted ­toward optometrists. Thus, in 2015, ophthalmologists spoke out against state legislation in Texas aimed at allowing optometrists to provide more ser­vices.27 Even two years l­ater, the Texas Ophthalmologist Association (TOA) website was blunt: “TOA ­will not sit idly by while a small, radical faction of non-­physicians seeks authority to treat patients with surgical and medical modalities in which they are not trained and which require a comprehensive medical education to fully manage.”28 The American Medical Association’s Historic Victories against the Public

The California Medical Association is a subsidiary of the AMA, which has a long and well-­documented history of limiting entry to markets to the advantage of its members and to the expense of the public. In 2016, only five organ­izations outspent the AMA on lobbying, three of which ­were healthcare related: Blue Cross/Blue Shield, an insurer, with close

Unequal Access to Professional Ser­vices  •  251

ties to medical associations; the American Hospital Association; and the Phar­ma­ceu­ti­cal Research and Manufacturers Association of Amer­i­ca. The AMA invested $19 million in lobbying, according to the Center for Responsive Politics.29 To appreciate how the AMA limits and restricts entry to markets for healthcare t­ oday, consider a few of its historic activities. As medical science advanced in the late nineteenth ­century, physicians gained newfound re­spect from the public and began to win out against less-­professional and less-­ organized providers of healthcare ser­vices. State licensing laws, which had been repealed for de­cades, w ­ ere reestablished ­a fter the Civil War in many states. In the early nineteenth ­century, orthodox medicine consisted of blood-­ letting and the inducement of mineral poisons. Homeopathic medicine and eclecticism—­ which recommended rest, herbal remedies, and baths—­ developed as competing models and established their own medical schools. The orthodoxy fought back through the creation of the American Medical Association in 1847, which nationally or­ga­nized state-­and university-­level organ­izations. Its goal, according to historian Ronald Hamowy, was largely to reduce the supply of competing schools and physicians.30 In 1867, the AMA made the introduction of state licensing restrictions a high priority through the establishment of AMA-­controlled licensing boards and eventually won enough control to deny licensure to gradu­ates from non-­A MA-­approved colleges, causing them to close down. AMA power to control institutions was particularly strong in the early twentieth c­ entury, and they used it to reduce the number of medical schools from 131 in 1910 to 88 in 1920 and 77 in 1939.31 Controlling How Healthcare Is Paid For

With the power to decide who could practice medicine established, the AMA went on to dictate how medical ser­vices w ­ ere paid for and by whom. In the 1920s and 1930s, medicine was becoming increasingly expensive, and many consumers could not afford it. Doctors had long respected the need to treat ­people who could not afford care and had a variety of pricing schemes in order to do so, which mostly involved having richer p­ eople pay more to subsidize the care of low-­income p­ eople. Still, as medical science deepened, research hospitals became more common, and specializations grew, doctors

252 • Chapter 9

figured out more efficient and effective ways to or­ga­nize healthcare to help the poor and improve the quality of care. In her power­ful book on the origins of U.S. health insurance, historian Christy Ford Chapin documents how u­ nions, large businesses, and other private organ­izations hired doctors on staff or other­wise pooled money from their members to collectively pay for certain medical expenses as needed.32 This group medical care model was a radical departure from the basic business model supported by the AMA in that it paid doctors a fixed salary—­from ­union or membership dues in some cases—­unrelated to the ser­vices performed, and thereby exerted much greater control over doctors than they ­were accustomed to. The basic model had been that doctors worked in­de­pen­dently at their own clinics and charged what­ever fees they deemed appropriate. Two examples illustrate the superiority of the cooperative or group model. In 1929, a Lebanese-­born doctor named Michael Shadid launched a healthcare cooperative in Elk City, Oklahoma, to better meet the needs of poor agricultural workers and other rural residents. For an entry fee ($50) to build and equip a hospital and a modest annual membership fee ($25), families had access to ­free clinical and hospital care.33 An annual fee of $25 in 1929 would have been 4 ­percent of the average annual income of Americans living at that time. As of 2018, f­ amily health insurance costs employers and workers an average of $19,600, which amounts to 37 ­percent of average income.34 Over 2,000 families signed up for Shadid’s plan and the model spread such that 101 rural cooperatives w ­ ere in place by 1950. Shadid published books about his philosophy of healthcare, emphasizing that group practice encouraged doctors to share knowledge, discuss cases, and practice preventive medicine, while creating an affordable way for patients to get treatment.35 Also in 1929, municipal workers in Los Angeles convinced Drs. Donald Ross and Clifford Loos to create a prepaid medical group for roughly 2,000 workers and their families. As Ford Chapin relates, the group created eventually included salaried doctors from dif­f er­ent specialties who could all work together to discuss individual cases and make contractual arrangements with hospitals for emergency cases. Thus, for a small fee, all of the patients’ healthcare needs could be met in an integrated system that did not require getting referrals from one in­de­pen­dent doctor to another and did not rely on arbitrary and unpredictable fee schedules. By 1935, the Ross-­Loos clinic had 50 employed doctors and 40,000 patients, a drugstore, a staff library, and a surgery office. Moreover, in additional to salary payments, the doctors ­were given a percentage of

Unequal Access to Professional Ser­vices  •  253

profits. Since revenue depended on attracting and retaining members—­and not on providing excessive services—­the doctors had a strong incentive to control expenditures by not providing unnecessary procedures, but like in­de­ pen­dent doctors, they also had a strong incentive to keep patients happy and healthy, so patients would remain paying members.36 The AMA opposed ­these models. At a time when corporations ­were gaining prominence and employing other professionals—­such as engineers—­ doctors did not like the idea of becoming mere employees ­under the control of some businessman in ­either a corporate or government bureaucracy. Although group practices ­were run by doctors, the doctors did not set the prices of treatments, and AMA leaders worried that groups of doctors would eventually be managed by outsiders if the principle of autonomy ­wasn’t held sacrosanct. In the 1930s, the AMA launched a systematic campaign to wipe out group practice and preserve the autonomous fee-­for-­service model. The AMA went so far to as expel Dr. Shadid from membership. This was a clear signal of where it stood on cooperatives, but the conspiracy was far more serious: The AMA lobbied aggressively and successfully to change state laws to make group cooperatives e­ ither illegal or effectively impossible to or­ga­nize.37 Similar tactics ­were used to crush group practice. In 1938, the AMA was indicted by a federal g­ rand jury for violating the Sherman Anti-­Trust Act. The AMA and its state affiliates, in this case the Medical Society of the District of Columbia and the Washington Acad­emy of Surgery, conspired to ostracize and coerce any doctors who formed such a group and restrain them from practicing.38 Ostracism by the AMA was eco­nom­ically fatal to doctors ­because, with few exceptions, only doctors with AMA membership w ­ ere allowed to treat patients in hospitals, and AMA members w ­ ere discouraged from consulting with and making or taking referrals to or from non-­A MA members. The net result was that group practice and cooperatives w ­ ere almost annihilated, while preserving physician autonomy to overcharge patients as they saw fit. The AMA’s hostility to group practice also had an unintended and deeply unfortunate result. Insurance companies and eventually, in the 1960s, the U.S. government began paying for healthcare to deal with the rising costs, but they had to do so according to the AMA’s rules, creating a fundamental conflict that still haunts the United States between ser­vice providers (prac­ti­ tion­ers) and payment providers (insurers).

254 • Chapter 9

Insurance companies w ­ ere very reluctant to enter the market, according to Ford Chapin, but ­were gradually convinced to do so. At first, they viewed health insurance as a losing proposition b­ ecause they had no way to control doctors—­who insisted on total autonomy and fee-­for-­service payments. Insurance com­pany concerns that fee-­for-­service payments would encourage the over-­provisioning of ser­vices have proven valid in the face of careful academic study.39 The health insurance model did begin to gain traction, however, as citizens and the government grew increasingly frustrated with the high fees charged by physicians at hospitals and in private practice, especially as reports of billing malfeasance and the use of unnecessary hospital admissions and surgical procedures became widespread. The insurance companies and federal bureaucracies in charge of health insurance responded by subjecting physician billing claims to intense scrutiny and audits. Physicians began spending hours per week on paperwork with the hope of using just the right codes and language to get approval for payment.40 Physicians’ offices and hospitals now routinely employee ­people who attend college and obtain certifications in the field of ­either medical coding—­ transforming a doctor’s notes into billable codes—or medical billing—­ entering codes into complex insurance-­company or government-­insurance rec­ords systems.41 According to the U.S. Bureau of ­Labor Statistics, over 200,000 U.S. adults worked in the occupation of “medical rec­ords and health information technician” as of 2017.42 Another 135,000 workers ­were classified as “billing and posting clerks” and worked in e­ ither physicians’ offices or hospitals.43 The same database reports 666,000 physicians and surgeons, which suggests that one billing professional is required for ­every two doctors.44 Current estimates suggest that U.S. consumers and taxpayers spend 4.6 times as much on healthcare administration as the average resident of a rich country (in the OECD).45 A detailed comparison between the United States and Canada found that U.S. nurses in physician’s offices spend an additional 18 hours per week pro­cessing billing codes and claims compared with their Canadian counter­parts and administrative staff spend an additional 27 hours per week.46 Healthcare economists estimate that U.S. residents waste hundreds of billions of dollars per year in administrative expenses that contribute nothing to the quality of care.47 Thus, the United States healthcare system exhibits two massive inefficiencies, which when combined, largely explain why U.S. residents spend around

Unequal Access to Professional Ser­vices  •  255

twice as much per capita on healthcare as residents of other rich countries without getting higher-­quality care: The AMA-­supported inefficiency of the fee-­for-­service system encourages physicians to provide excessive ser­vices, while the inefficiency of the claims and billing system forces physicians to waste massive resources on administrative paperwork. The latter is deemed a necessary evil to counteract the bad incentives created by the former. Both prob­lems would have been avoidable if the AMA had allowed cooperatives and group practice to flourish. Unfortunately, the AMA viewed such organ­ izations as competitive threats, and rather than rise to the challenge, the AMA hobbled its competition, with disastrous implications for the country, which now spends 18 cents of e­ very dollar on healthcare ser­vices.48 Restraining group practice in f­ avor of a fee-­for-­service insurance model is only one way that the AMA has controlled the development of U.S. healthcare. Another crucial way in which the physician lobby has harmed the public and exacerbated inefficiency in healthcare delivery is by insisting that doctors cannot be employees or even in­de­pen­dent contractors of nondoctors. This set of laws is called the corporate practice of medicine. In nearly e­ very state—­including the largest ones of California, Florida, Texas, and New York—it is against the law for a nondoctor to manage a business that employs doctors.49 Thus, a brilliant software designer or business management expert with ideas about how to provide healthcare more efficiently is forbidden from raising venture capital money, starting a new model, and scaling it up if it involves the employment of physicians. The only common exception is that hospitals and medical universities are allowed to be managed by nondoctors in many states. The law is justified as a means of removing financial incentives from medical decision-­making, but as ­we’ve seen, the current model essentially encourages physicians to overcharge patients and insurers, and the AMA has a long history of restricting entry into markets to benefit physicians. Medical Associations in Other Developed Countries

As noted, healthcare is inordinately expensive in the United States, and while doctors are well paid in e­ very country, they are particularly well paid in the United States, making them the largest source of top-1-­percent income earners by occupation—­even more than chief executives. Other countries have medical lobbies, but they have not been as influential in steering policies to ­favor physicians as the AMA.

256 • Chapter 9

While the educational requirements for ­those who wish to become physicians are similar across developed countries, they tend to require fewer years of training outside the United States. Many developed countries do not require that someone first earn a bachelor’s degree to enter medical school, as is the case in the United States. Instead, students in Australia, Canada, Ireland, the UK, and the Netherlands can enter medical school straight from secondary school, though that route requires an extra two years compared to t­ hose with a four-­year degree. On average, that saves two years of formal college instruction. In Italy, the Netherlands, and Canada physicians are not required to spend as many years in post–­medical school training, saving what amounts to roughly two years, depending on the specialty.50 Similarly, countries outside the United States are often more favorably disposed to professions that compete with physicians. In Canada nurse prac­ti­tion­ers practice in­de­pen­dently without any need for supervision from physicians.51 The same is true in Australia and the UK, despite opposition from the Australian Medical Association and the British Medical Association.52 In a 1960 book about the British Medical Association (BMA), which is or­ga­nized like the AMA in many re­spects but has been far less po­liti­cal, the late Harry Eckstein, a po­liti­cal scientist, wrote: The plain fact is that the BMA has lost all the crucial public disputes over medical policy in which it has engaged during this ­century. Twice the government has passed over BMA opposition legislation affecting the ­whole structure of medicine—­opposition which was intense despite the broad area of agreement between the profession and the government. From this standpoint, the BMA emerges as a po­liti­cal group considerably weaker than the American Medical Association.53 It is unclear why the AMA has been so much more effective than its counter­parts. One ­factor may be the decentralized power of American market regulation, which largely rests in the hands of states, not the central government. Doctors and ­lawyers provide local ser­vices and tend to be diffused across cities and counties, and they rarely have to contend with other power­ ful groups at the local and even state level, whereas pro-­consumer and pro-­ labor interest groups can often successfully mobilize to influence federal

Unequal Access to Professional Ser­vices  •  257

regulations. The AMA has blocked federal legislation to expand government healthcare, but in my opinion, as discussed above, its most blatant suppression of competition occurs at the state level.

DENTAL SER­V ICES

Dentists operating in the United States enjoy income advantages that are almost as large as ­those of physicians. From 2012 to 2016, 19 ­percent of dentists w ­ ere earning incomes that put them in the top one p­ ercent. That makes dentists second only to physicians and surgeons in terms of the likelihood of being in the top one p­ ercent, among 480 detailed occupational categories. The high wage and salary income translates into substantial wealth. The average home value of dentists is $638,000, which puts them ahead of chief executives and legislators ($621,000). It is not unusual for countries to impose limitations on the supply of dentists; nor is it unusual for dentist associations to control or at least heavi­ly influence this pro­cess. In the Netherlands, for example, the Royal Dutch Dentistry Association helps determine the national quota for the number of students admitted to dental school. As a result, the number of ­people graduating with the ability to practice dentistry has barely changed over the past three de­cades. ­There is a relatively small number of dentists in the Netherlands compared to its overall population, and a large number of p­ eople who apply for dental school never gain admission. The Dutch Health Care Authority sets maximum fees that dentists can charge, but they nonetheless enjoy a major earnings advantage—of about $68,000, or 55,000 euros per year—­ over ­people who work similar hours and who have similar cognitive per­for­ mance on a nationwide secondary exam, according to Dutch economists.54 In the United States and elsewhere the high earning premium for dentists gives the public strong incentives to shift tasks to other professionals who could provide similar ser­vices at a lower cost. Dental hygienists play that role to a large extent, and they require fewer years of postsecondary education. Not surprisingly, the American Dental Association has aggressively lobbied to block dental hygienists from opening their own clinics and practicing without their supervision.55 As a result, ­there are many ser­vices that dental hygienists are not allowed to perform despite possessing the requisite training, or that they can perform only u­ nder vari­ous levels of supervision.56 For example, in 41 states and the District of Columbia, dental hygienists cannot

258 • Chapter 9

administer local anesthesia without a dentist on site, and in Texas, they cannot administer u­ nder any circumstances.57 Only a small number of states permit dental hygienists to provide ser­vices to patients without supervision by a dentist, even in schools, hospitals, and long-­term care facilities, where having a dentist on staff would be prohibitively expensive. Seeing a need for lower-­cost access to dental ser­vices, Alaska, Minnesota, and Maine have created a new dental therapist license, which permits dental professionals with similar training to that of a dental hygienist to provide basic preventative dental care. With pioneering efforts credited to New Zealand, one 2012 report finds that 54 countries use dental therapists as auxiliary providers of oral healthcare, usually in public school settings.58 Despite the obvious and well-­documented public health benefits of such programs, the American Dental Association has stridently opposed efforts to grant ­legal status to ­these prac­ti­tion­ers.59 One reason why dentists have successfully blocked their competitors is that they are given broad powers to decide what constitutes the practice of dentistry and prevent nondentists from performing t­ hose activities. In one example, the Federal Trade Commission—­which is charged with protecting the public from uncompetitive business practices—­successfully won a case in the U.S. Supreme Court against the North Carolina Board of Dental Examiners. The board was stacked with practicing dentists—as most licensing boards are—­and used that power to crack down on the provision of teeth-­ whitening ser­vices by nondentists ­after dentists complained to the board about losing business to more affordable competitors.60

­L EGAL SER­V ICES

Given the training and skill requirements, it is not surprising that l­ awyers tend to receive relatively high compensation compared to other professions. Yet, as with doctors, dentists, and financial-­sector workers, both straightforward evidence (examining pay as it relates to education or cognitive ability) and advanced statistical analy­sis (controlling for even unobserved skills) show that l­ awyers are overpaid, even ­after accounting for their skills.61 Comparable international data on l­awyers is difficult to find, but the Luxembourg Income Study provides information on incomes for ­lawyers in a handful of countries and further provides earnings in a number of countries for “­legal, social, and cultural professionals,” which includes ­lawyers,

Unequal Access to Professional Ser­vices  •  259

librarians, social scientists, authors, and journalists.62 ­These data indicate that ­lawyers and related professionals are often highly likely to be top income earners and comprise a substantial share of top income earners in many countries. In Luxembourg, l­egal, social, and cultural professionals comprise 18.2  ­percent of top earners. In the United States, the narrower category of ­lawyers and judges comprises 7.7 ­percent, which is the highest in the Luxembourg Database for narrowly defined ­lawyers. Indeed, the probability of a ­lawyer being a top-1-­percent income earner is higher in the United States (at 14.2 ­percent) than in any other country that reports ­lawyers narrowly. Though lower, the top-1-­percent rate is also high in Germany and Denmark, where 12 ­percent and 10 ­percent, respectively, of ­lawyers are top earners. Yet, it is not inevitable that ­lawyers are concentrated at the top of the income distribution. In the Netherlands, Finland, and Switzerland, very few top earners (5 ­percent or less) are in the broad category of “­legal, social, and cultural professionals.” This variation suggests that country-­specific institutions and po­liti­cal ­factors shape the compensation of l­ awyers (­table 9.3). An OECD report on professional ser­vice competition notes that ­lawyers benefit from a mono­poly on representing clients in court in almost e­ very OECD country, and in many, they also have a mono­poly on providing paid ­legal advice or applying for patents. In some countries, l­awyers are required even to make routine real estate transactions or deal with the allocation of assets following the death of a ­family member.63 Notably, Finland and Sweden have the least restrictions on the provision of ­legal ser­vices among developed countries. In Sweden, nonlawyers can represent defendants in court, and that was also true in Finland u­ ntil 2002.64 In both countries, a license is not required to provide l­ egal advice, tax advice, or patent filing, which is also the case in the Netherlands. In the United Kingdom, Belgium, Ireland, and Italy nonlawyers may provide tax advice. In Austria, Denmark, France, Germany, Greece, Luxembourg, Portugal, Spain, and the United States, only l­ awyers can represent p­ eople in court or provide ­legal advice.65 Notably, l­ awyers appear to be less prevalent among top income earners in Finland and the Netherlands, where markets are more open, compared to the United States, Luxembourg, Denmark, Germany, and Spain. The Control of ­Legal Ser­vices in the United States

The American Bar Association (ABA) was developed, in part, with the worthy aim of raising the quality of l­ egal ser­vices in the United States, but from

260 • Chapter 9

­ able 9.3.  Probability of lawyers being in top 1 percent of income T earners, by country

Country

Title

Luxembourg

­Legal, social, cultural professions ­Lawyers and judges ­Lawyers ­Lawyers ­Lawyers ­Lawyers ­Legal, social, cultural professions ­Legal, social, cultural professions ­Lawyers ­Lawyers ­Legal, social, cultural professions ­Legal, social, cultural professions ­Legal, social, cultural professions ­Legal, social, cultural professions ­Legal, social, cultural professions ­Legal, social, cultural professions ­Legal, social, cultural professions

United States Germany Denmark Estonia Uruguay Poland Spain Israel Rus­sia Ireland Czech Republic Greece Switzerland Austria Finland Netherlands

Year of data collection

­ ercentage P of ­lawyers who are top-1% earners

­ ercentage P of top-1% earners who are ­lawyers

2013

14.3%

18.2%

2012–2016 2013 2013 2013 2013 2013

14.2% 12.1% 9.7% 7.1% 5.7% 5.7%

7.7% 6.6% 2.5% 2.0% 2.9% 8.8%

2013

5.6%

13.6%

2012 2013 2010

5.0% 4.7% 2.6%

7.8% 4.9% 9.4%

2013

2.1%

3.3%

2013

2.1%

5.4%

2013

1.4%

4.6%

2013

1.2%

3.8%

2013

0.8%

5.1%

2013

0.8%

5.3%

Sources: All data is from the Luxembourg Income Study, except for the United States. ­Those data are from IPUMS USA. Luxembourg Income Study (LIS) Database, http://­w ww​.­lisdatacenter​.­org (multiple countries; run May 24, 2017–­June 17, 2017). Luxembourg: LIS.

its beginning it also had the destructive and harmful goal of raising the cost of becoming a ­lawyer and limiting competition for existing ­lawyers. The second goal is in direct conflict with the public good but highly valuable to ­lawyers, and so the ABA’s efforts have resulted in unreasonable limitations on ­legal practice. The seductive power of the second goal has, in fact, over-

Unequal Access to Professional Ser­vices  •  261

whelmed even the first. ­Little effort is made by the ABA to provide the public with information about the quality of l­ egal ser­vices or even the quality of law schools. If the ABA ­really wanted students to get the best-pos­si­ble ­legal training, it would publish a user-­friendly database that would allow ­people to assess the quality of law schools. This could entail releasing data at the institutional level on bar-­exam passage rates by state as well as more sophisticated metrics that adjust for the difficulty of state exams and take into account differences in test scores and income of entering students. It could even collect and report data from alumni surveys, with basic questions on a Likert scale such as “How would you rate the quality of your l­ egal education?” or “How likely would you be to recommend your law school to ­others?” If the ABA wanted the public to employ the best l­ awyers, it would create a website that would allow an individual to more readily find a l­ awyer who is the right match for him or her. Nothing is stopping the ABA from creating a searchable website that allows the public to find licensed l­awyers by their specialty, discover when they obtained their license, and maybe even where they attended law school. L ­ awyers could even enter resume-­like information highlighting their accomplishments and experience—­even their hourly rates. Of course, the ABA does not do any of this—­nor does the AMA. By contrast, if you want to hire a domestic worker, companies such as Care​ .­com allow you to make ­these sorts of comparisons. If you want a designer, you can use something like 99 Designs to have designers bid for your proj­ect and submit prototypes. Websites such as Upwork are designed for a broad range of professionals—­including accountants and software developers. You ­will not, however, find l­ awyers and doctors on ­these websites. Instead, the ABA focuses its considerable resources and energy on ensuring that the public continues to transfer a large and excessive share of income to ­lawyers. It does this in several ways. First, the ABA’s actions support a high barrier to entering the profession. To practice law one must gradu­ate from an accredited law school and pass a state-­specific bar exam that usually does not allow one to practice outside that state. In 2015, 59 ­percent of exam takers passed in the average state.66 Aside from the anti-­competitive issues raised by the bar exam, the Society of American Law Teachers points out that it is deeply flawed in several other dimensions. According to the society, the test mea­sures one’s ability to memorize a wide swath of ­legal princi­ples and procedures, but it does not test the skills actually used by ­lawyers, which include research. It would be considered malpractice to give a client advice based solely

262 • Chapter 9

on memory, without performing ­legal research, but the test does not demand competency in this area.67 Unsurprisingly, passage rates are lower for lowerincome groups, who are less likely to be able to afford a bar exam class or devote many hours to preparation for the exam. Second, the ABA defines the practice of law in an expansive way such that even mundane ser­vices require a licensed ­legal attorney. For example, some states actually require an attorney to be physically pre­sent when the deed of ah ­ ouse is transferred, whereas ­others require attorneys to prepare the documentation.68 In any case, this raises the costs of “closing” on a ­house significantly, regardless of the fact that the tasks involved require very specific bureaucratic knowledge but in no way require a comprehensive grasp of American ­legal ­matters—­much less the passage of a bar exam. If it ­were up to the ABA, the situation would be even worse, ­because not only would such ser­ vices require an attorney, they would require high and fixed prices. In 1975, plaintiffs purchasing a home in ­Virginia won a lawsuit against the ­Virginia Bar Association for inflating the price of title examinations. The Bar Association had been publishing pricing guidelines for such ser­vices and disciplining members if they did not adhere to ­those prices.69 Third, the ABA, like many of its international counter­parts, prevents nonlawyers from operating l­ egal ser­vices firms. The American Bar Association’s Model Rules of Professional Conduct prohibit nonlawyers from owning or even managing law firms, as well as preventing l­awyers from sharing fees with non­lawyers.70 This is the law prevailing in ­every state, though not in the District of Columbia, where nonlawyers may have an owner­ship stake in a ­legal ser­vices firm.71 Outside of the United States, Australia has allowed nonlawyers to own law firms since 2001, and ­England and Wales since 2007. Other Eu­ro­pean countries allow nonlawyers to own a minority stake.72 As with rules barring the corporate practice of medicine, the stated goal is to preserve the integrity and in­de­pen­dence of l­egal judgment and advice, but, like the corporate practice of medicine prohibitions, the idea that the practice of law would be corrupted if ­lawyers had to work for a nonlawyer is specious. Law firms owned by ­lawyers face just as much pressure to make money as law firms owned by entrepreneurs who are not l­awyers. In effect, the rule ensures that owner­ship profits related to ­legal ser­vices are appropriated exclusively by ­lawyers, and it mitigates ­potential downward pressure on fees that more efficient business organ­i­ zations might create.

Unequal Access to Professional Ser­vices  •  263

Unmet L ­ egal Needs

The American public does not seem to think highly of ­lawyers. Since 1976, Gallup has asked a random sample of U.S. adults to say how they would rate “the honesty and ethical standards of ­people in t­ hese fields.”73 ­Lawyers are consistently ranked near the bottom. The highest percentage of U.S. adults who answer “high” or “very high” with re­spect to l­ awyers has never been more than 27 ­percent; it stood at 19 ­percent in the latest poll from 2018. By contrast, 84 ­percent of U.S. adults rate the honesty and ethical standards of nurses as at least “high,” and roughly two out of three give that rating to medical doctors, pharmacists, military officers, and teachers at high ­school and grade ­school levels. Even with the high-­profile controversies over policing discussed in the previous chapter, 54 ­percent of U.S. adults rate police officers as having high or very high honesty and ethical standards. This negative reputation of ­lawyers is unfortunate ­because ­legal ser­vices are massively impor­tant and often under-­sought and under-­provided. The ­legal community has long known that low-­and moderate-­income individuals are particularly likely to have unmet civil l­ egal needs. L ­ egal needs, in this sense, are defined as noncriminal situations that involve pos­si­ble violations of rights or responsibilities recognized by laws, where a person might benefit from l­ egal advice or action. A major survey by the American Bar Association in 1992 concluded that approximately half of low-­and moderate-­income U.S. residents faced a noncriminal (or “civil”) ­legal need during the year.74 The most-common ­legal needs involved consumer financial prob­lems, such as prob­lems with creditors, taxing authorities, or insurance companies or inability to obtain credit or tax; housing prob­lems, such as unsafe or problematic living conditions; and domestic prob­lems, such as divorce mediation or disputes over childcare support. Seventy-­one ­percent of low-­income ­house­holds and 61 ­percent of moderate-­income h ­ ouse­holds with l­egal needs did not take formal ­legal action, citing cost concerns as one of the most impor­tant reasons. More recently, in 2003, Washington State published a landmark study with an improved methodology, which found that roughly 80 ­percent of low-­ income Washington ­house­holds experienced at least one situation that would call for noncriminal l­ egal counsel within a year; most had multiple needs. In all but 15 ­percent of cases, low-­income p­ eople confronted their noncriminal ­legal challenges without help from an attorney, according to the

264 • Chapter 9

study.75 A 2015 follow-up study found similar results.76 Efforts to mea­sure unmet civil l­ egal needs in the UK have reached similar conclusions.77 Additionally, the functioning of the criminal justice system is greatly hindered by inadequate funding. Most criminals cannot afford attorneys, and the public defenders—to whom they are entitled—­are massively over­ burdened, understaffed, and underpaid, as the ABA and U.S. government have long admitted.78 ­Simple and routine cases drag on for years, holding even innocent p­ eople’s lives in forced abeyance, while delaying justice for both victims and guilty parties. Though the prob­lem is national, Washington is arguably the only U.S. state to take substantial action to tackle this issue. First of all, it is one of a small number of states that allows nonattorneys to prepare and complete ­legal documents related to real estate transactions. Since 1983, the Washington State Supreme Court has allowed “­limited practice officers” to do this, ­after passing an exam and obtaining the appropriate license, whereas most states require a licensed attorney.79 Since then, it has taken even more radical action, with far-­reaching implications, but the opposition to the court reveals the power of professional elites to preserve their economic privileges and explains why t­ hese changes have not occurred elsewhere in the United States. In 2001, the Washington State Supreme Court created the Practice of Law Board to investigate the unauthorized practice of law, advise the state court and l­egal community, and recommend ways to increase access to l­egal ser­ vices.80 The board was administered and funded by the Washington State Bar Association, and in 2006 it proposed the creation of ­legal technicians who could provide some ­limited ­legal ser­vices, ­after completing at least an associate’s degree in relevant l­ egal studies, passing vari­ous exams, and working u­ nder the supervision of a l­awyer.81 The proposal was rejected by the Washington State Bar Association’s board of governors in 2006 and again in 2008, ­after revisions. Perhaps as a result, the state Supreme Court repeatedly delayed votes. Many ­lawyers in Washington and California, which was also considering a similar policy, wrote letters or other­wise publicly criticized the proposal. The criticisms reveal the low regard that elite professionals have for ­those who have not gone through their same level of training: “The class of individuals described is not ­going to have the competency to actually do for the poor what needs to be done,” one partner in a Washington law firm argued. “The state bar proposal now ­under consideration would simply give the veneer of legality

Unequal Access to Professional Ser­vices  •  265

to t­hese unauthorized, ill-­trained prac­ti­tion­ers who do more damage than good,” argued a California ­lawyer about a similar proposal in his state.82 Fi­nally, in 2012, the Washington State Supreme Court ­adopted the recommendation and created a L ­ imited License ­Legal Technician (LLLT), who is licensed to provide l­ egal counsel and prepare l­ egal documents but not represent clients in court or negotiate on their behalf.83 Currently the license is applicable only to domestic ­legal affairs, such as divorce and child support and custody issues, but the plan is to gradually expand the scope. Importantly, the LLLT’s educational requirements are far more streamlined, and similar to those of a para­legal, with a few additional steps. In the Seattle, Washington, metropolitan area, the average l­ awyer earns a salary of $139,000, compared to $64,200 for a para­legal, so the LLLT license has the power to overcome many of the cost barriers that currently deter ­people from seeking ­legal assistance.84 Divorce mediation, to offer but one example, is an impor­tant l­egal ser­vice that an LLLT can provide and save a ­house­hold thousands of dollars. The mediator serves an impor­tant function by guiding a divorcing c­ ouple through the vari­ous issues they are likely to confront, such as childcare arrangements and the allocation of financial assets and responsibilities. The mediator helps them reach a solution that is mutually aggreeable and able to withstand scrutiny from a judge in divorce court. ­These issues can be complex and contentious, but they do not require years of case study or courtroom experience. Rather, the issues require someone who is skilled at understanding and addressing p­ eople’s concerns, walking them through vari­ous options, and condensing the final information into a fairly standardized and routine document. ­A fter its members proposed the LLLT to the Supreme Court, the state bar association’s Board of Governors reportedly voted to defund and eliminate the Practice of Law Board, according to the Practice of Law Board’s former members.85 In an open letter to the Washington State Supreme Court, the chairman and three other board members resigned in protest, stating that “the Washington State Bar Association has a long rec­ord of opposing efforts that threaten to undermine its mono­poly on the delivery of l­ egal ser­vices.”86 The LLLT program is not old enough and the number of licensees not large enough for the sort of detailed studies that compare nurse prac­ti­tion­ers to physicians, but qualitative evidence gathered by scholars at the National Center for State Courts and the American Bar Foundation contradicts the elitist concerns raised by l­ awyers opposed to the license. The authors interviewed

266 • Chapter 9

13 of the 15 LLLTs who had been licensed and several of their clients and colleagues. They concluded that “clients uniformly reported that the LLLTs provided competent assistance” and that their “­legal outcomes ­were improved by utilizing the ser­vices of the LLLTs.” The authors recommended that the program be expanded to other states.87 Roughly a year ­a fter that report was published, the number of LLLTs doubled in Washington, where t­ here are presently 30 ­people with LLLT licenses in Washington, according to the state bar association directory.88

FINANCIAL SER­V ICES RESERVED FOR THE RICH

Any study of income in­equality in the United States must take into account the role of the financial sector and the fact that most of the inordinate income gains go to t­ hose working in securities, commodity contracts, and investments. In 2016, average compensation for all workers in the United States was $75,000, but more than three times that amount for workers in the securities and investment industry ($255,000). That makes it by far the highest-­paid industry.89 Compensation is even well ahead of other high-­paying industries such as oil and gas ($202,000), as well as tech industries, including information and data-­processing ser­vices ($155,000), software and other publishing ($149,000), computer and electronics manufacturing ($138,000), and professional, scientific, and technical ser­vices ($111,000). In fact, workers in the securities and investment industry are paid more than double what workers in banking are paid, which is just $97,000. Data from the Census Bureau from 2012 to 2016 shows that the securities and investment industry had the highest percentage of workers in the top one ­percent (10.5 ­percent); that is notably higher than the medical (7.8 ­percent) and l­ egal professionals (7.7 ­percent), which ­were the second-­and third-­highest paying, respectively, although ­those two accounted for a larger absolute number of top earners as a result of their larger base of workers.90 It is also worth noting that workers in the consumer banking industry are rarely in the top one ­percent (only 2.4 ­percent). Investment banks and hedge funds make up the majority of businesses in the securities and investment industry. At Goldman Sachs, one of the highest-­ paying firms in the industry, average compensation per worker in 2016 was

Unequal Access to Professional Ser­vices  •  267

$339,000.91 But even Goldman’s extraordinary compensation falls below estimates for the median hedge fund, where median salaries ­after bonus payments amount to $350,000.92 The average would be substantially higher. In fact, large investment banks w ­ ere increasingly integrated with hedge funds ­until the regulatory changes that followed the G ­ reat Recession made it difficult for firms such as Goldman Sachs to operate hedge funds. Now, Goldman collects fees to refer clients to hedge funds, often t­ hose managed by former Goldman employees who broke away a­ fter the regulatory changes.93 I have come to believe that ­these extraordinary salaries exist only ­because of very specific regulations that both shelter hedge funds fees from market competition and prevent mutual funds from using the same strategies.94 Skill alone d­ oesn’t explain the high salaries, ­because workers with the same or higher levels of skill in other industries are paid less. Instead, ­there are good reasons to believe that hedge funds and investment banks are operating in a market that is fundamentally not competitive.95 To be sure, like doctors and ­lawyers, workers at investment banks—­ including JPMorgan, Goldman Sachs, Morgan Stanley, and Credit Suisse—­ possess specialized and valuable skills. They provide extremely complex ser­vices to large, sophisticated clients, such as multinational corporations, start-­ups in the midst of an initial public offering, massive pension funds, university endowments, and governments. The richest and largest clients of investment banks genuinely benefit from the expertise that ­these firms possess. The fact that financial-­sector workers who serve large organ­izations earn relatively high salaries compared to workers with no specialized knowledge or education is not itself a scandal. The prob­lem is that the size of the salary advantage is disproportionate to the value of the under­lying skills and thus serves as a tax on the economy, as I w ­ ill explain. In analyzing finance, it is impor­tant to distinguish between the activities that serve institutions and rich investors and t­ hose that serve ordinary individuals. In contrast to investment banks, the big retail banks do not make vast profits from the basic routine transactions that retail consumers conduct on a regular basis—­such as getting a home loan, refinancing a mortgage, buying and selling stocks, or maintaining a checking account. ­Those are relatively low-­profit-­margin activities. With appropriate regulation and oversight to protect against fraud and abuse—­which was missing during the housing ­bubble—­people can readily get ­these financial products at a reasonably low cost. Evidence for this comes from the U.S. Bureau of Economic Analy­sis

268 • Chapter 9

­ able 9.4.  Change in prices of vari­ous goods and ser­vices, 1980–2016, T in the United States Price index 2016/ Price index 1980 All goods and ser­v ices

2.5

Goods

1.5

Electronic equipment and media Motor vehicles and parts Clothing and footwear

0 1.7 1.1

Ser ­v ices

3.4

Housing Healthcare Financial ser ­vices Communication Higher education ­Legal, accounting, and other professional ser­vices

3.6 4.8 1.8 1.4 11.4 5.7

Source: U.S. Bureau of Economic Analy­sis, T ­ able 2.4.4. Price Indexes for Personal Consumption Expenditures by Type of Product.

(­table 9.4). The price of financial ser­vices has been rising more slowly than overall inflation—in sharp contrast to other ser­vices such as housing, healthcare, education, and ­legal ser­vices. Financial ser­vices prices have faced downward pressure for two reasons: The first is information technology—­including automated teller machines and online banking, which have streamlined access, increased competition, and lowered costs; the second is the deregulation of banking within and across states, which occurred largely from the 1970s through the 1990s. Up u­ ntil the 1970s, most states prohibited banks headquartered in other states from opening branches. One reason for this is that state taxing authorities could charge charter fees and often owned part of the banks, and so competition from banks from other states weakened their tax base. States also prohibited branching within their own state, so that their preferred banks could benefit from mono­poly prices. Technological changes—­including telephone banking—­allowed customers to bank outside their local areas, and this and information technology helped create pressure for reform.96 The economics lit­er­a­ture generally finds that t­ hese regulatory reforms benefited economic per­for­mance by improving the quality of banking ser­vices.97

Hedge-fund performance fees (right axis) Hedge-fund management fees (left)

19.6 19.5 19.4 19.3 19.2 19.1 19.0 18.9 18.8 18.7 18.6

Hedge-fund performance fees

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Management fees (mutual and hedge funds)

Unequal Access to Professional Ser­vices  •  269

Mutual-fund fees (left)

Figure 9.1. ​Average mutual-fund versus hedge-fund fees, 1996–2016 Sources: Mutual-fund fees expressed as expense ratio for long-­term funds using data from Collins, Sean, and James Duvall. 2017. “Trends in the Expenses and Fees of Funds, 2016.” ICI Research Perspective 23, no. 3 (May). Available at www​.­ici​.­org​/­pdf/per23-03.pdf. Hedge-fund fee data are from Preqin Hedge Fund Online. Available at http://­docs​.­preqin​.­com​/­newsletters​/­hf​/­Preqin​-­HFSL​ -­March​-­2017​-­Hedge​-­Fund​-­Fees​.­pdf.

Beyond retail banking, retail investment fees have plummeted. In 1996, for ­every $1,000 the average person invested in a stock-­based mutual fund, he or she had to pay 1 ­percent each year, or $10. That’s not outrageously high but adds up over time. With the rise of low-­cost funds such as Vanguard—­ which was started in 1975 by John Bogle, a champion of the common investor—­and other index funds with l­ittle or no active management, that fee steadily declined. As of 2016, the average mutual fund takes just .63 ­percent, or $6.30 (figure 9.1). Moreover, for a Vanguard fund that gives a return based on the 500 largest publicly traded companies in the United States, the cost is just 0.04 ­percent, or $0.40 per year.98 Remarkably, nothing like this has occurred for hedge funds. Fees are roughly the same now as they ­were when modern hedge funds ­were in­ven­ted in the 1950s and are ­orders of magnitude higher than fees for mutual funds, ­because hedge funds take a much larger percentage of an initial investment (close to 2 ­percent) and a massive share of the profits from investment (close to 20 ­percent). The first modern hedge fund was credited to Alfred Winslow Jones, a former financial journalist turned investor, who got the idea in 1949. He

270 • Chapter 9

charged his investors 20 ­percent of profits, which ended up being substantial.99 Mutual funds are not even allowed to charge a per­for­mance fee. How is it pos­si­ble that mutual funds are so much cheaper than hedge funds? The answer lies in the only meaningful difference between the two. A fund, first of all, is an investment portfolio managed by a firm. “Hedge fund” is not a ­legal term, but refers to a fund that is less regulated than other funds, with the core distinction being that only rich individuals—­those with $1 million in assets or an annual income of $200,000 or more for at least two years—or institutional investors (e.g., pension funds, university endowments) can invest.100 ­These eligible investors are known legally as accredited investors, and in addition to hedge funds, they have access to private equity and venture capital markets that are not available to t­ hose who are not wealthy. This means they can buy stocks of companies that are not publicly traded or buy stocks before a com­pany goes public or gets acquired. Moreover, private funds—­hedge funds or other­wise—­that explic­itly refuse money from anyone except the very rich are automatically exempt from most securities regulations. Th ­ ese investors—­called qualified purchasers—­ must have a net worth of $5 million or more.101 Indeed, ­these investors are preferred by hedge funds ­because hedge funds are allowed to charge them per­ for­mance fees, not just management fees.102 So, a hedge fund is like a normal fund, except it can do almost anything it wants and is available only to rich ­people or large investment organ­izations. This regulatory advantage allows hedge funds to take more risk and use a variety of strategies unavailable to mutual funds. The main advantage is the use of leverage, which means using borrowed money and in some cases borrowed money that greatly exceeds the firm’s assets by many multiples. This is a huge advantage, which is why hedge funds outperform traditional funds.103 A large study of global hedge funds from 1994 to 2011 calculated the yearly average return on investment as just over 9 ­percent, ­after fees, compared with 7 ­percent for stocks and 6 ­percent for bonds.104 For large investments, two to three percentage points means millions if not billions of dollars. And for an individual investor, that premium is also substantial, even with the large fee. Leverage—­the ability to borrow in order to invest—is so advantageous ­because paying back a loan is cheap compared to a g­ reat investment strategy. Say you have you a good investment strategy that can easily scale into the millions—­buying or shorting commodities, for example—­and you expect a return of 20 ­percent. Let’s compare a hedge fund to a mutual fund for a fund with $1 million in investments from individuals.

Unequal Access to Professional Ser­vices  •  271

The mutual fund uses exactly the same strategy as the hedge fund but c­ an’t borrow, so its return is simply 20 ­percent of $1 million, or $200,000. The mutual fund earns 1 ­percent from management fees for a profit of $10,000. The hedge fund, however, gets a loan from some big bank for $100 million at 3 ­percent interest. At the end of the year, it pays the bank $3 million, but has turned the $1 million investment into $17.2 million in gains. The hedge fund man­ag­ers get $20,000 in management fees plus $3.4 million in per­for­mance fees. That’s why hedge fund man­ag­ers are so rich. So, why d­ oesn’t competition among hedge funds drive down the price of management fees? I believe ­there are two reasons. First, the Securities and Exchange Commission (SEC) made it illegal for hedge funds to advertise before 2013.105 The implication is that rich p­ eople and institutional investors had a hard time finding “better deals.” ­There was no reason for any one fund to lower its prices, ­because the anti-­advertising rule would have blocked it from using its lower prices to attract more investors. The SEC justified the rule as a way to protect consumers, but the commission made a huge m ­ istake. The prevention of advertising is r­ eally a way to keep consumers in the dark and encourage price collusion. The mutual fund price drop shows what could have happened. Mutual funds fees are public knowledge and easily found, and companies such as Vanguard advertise t­ hose low fees regularly. The second reason that hedge funds fees have remained so high is that most of the payments of t­ hose fees are indirect, from institutional investors. Institutional investors manage money from ordinary ­people, such as ­union workers and vari­ous retirement funds. Data from the SEC show that pension funds are the largest o­ wners of hedge funds.106 If one excludes other private funds—­ which include hedge funds and are also owned primarily by pension funds—­ and unclassified investors, over two-­thirds of hedge-­fund owner­ship comes from the following: private pension funds (such as ­those collected for u­ nion members), state or municipal pension funds, sovereign wealth funds, SEC-­ regulated funds (such as mutual funds), or nonprofits, such as university endowments. The man­ag­ers of ­these funds have very l­ ittle incentive to push down fees. Using SEC data, I estimate that only 20 ­percent of the owner­ship stake in hedge funds is attributable to individuals. Institutional investors who manage pension funds cannot charge their clients fees based on per­for­mance; their revenue is based on how much money they manage. Thus, they have almost no incentive to negotiate with hedge funds for lower per­for­mance or management fees, u­ nless their client—­a ­union

272 • Chapter 9

or a state government bureaucrat—­threatens to replace them for poor per­ for­mance. Since hedge funds often “beat the market,” no one seems to notice or care if huge percentages are siphoned away. In short, the prob­lem is that institutional investors are managing other ­people’s money with no “skin in the game,” to borrow a relevant phrase from Nassim Taleb.107 Just as healthcare consumers have ­little incentive to haggle with their doctor over the cost of a surgery paid for by an insurer, so a pension-­ fund man­ag­er has no incentive to tell a hedge-­fund man­ag­er to stop unnecessarily gouging his clients so he can buy another painting from the Italian Re­nais­sance for his summer mansion. As of 2017, private, state, and municipal pensions had roughly $686 billion invested in hedge funds, according to SEC data. If the hedge funds earned a 10 ­percent return, which is near the historical average, the pension plans would have given the hedge funds $161 billion over the next year at average prevailing fees.108 By contrast, if the pension plans had negotiated the fees down to mutual-­fund levels, they would have had to pay only $4 billion over the year. Say hedge-fund per­for­mance fees could be negotiated down to 5 ­percent (from 19.5 ­percent) with the same elevated 1.5 ­percent management fee. That would result in roughly $100 billion in annual savings for pension holders and would represent a massive re­distribution of capital income from elite professionals to ordinary workers. Mutual funds could never get away with ­these expenses—­even if they could charge per­for­mance fees—­because individual “retail” investors are often heavi­ly involved in picking them and the prices and fees are transparent, even openly advertised. The bottom line is that the SEC’s au­then­tic but misplaced concern for ordinary Americans has resulted in the w ­ holesale transfer of trillions of dollars from plumbers, pipefitters, teachers, automakers, and families with retirement accounts to super-­elite billionaires. Rich investors, like other elite professionals, are given privileged access to markets. In sum, then, the largest and richest groups among elite income earners in the United States benefit greatly from specific unjust laws. This po­liti­cal arrangement results in the mundane re­distribution of money from the bottom 99 ­percent of income earners to the top one ­percent and is unrelated to the ­things that scholars often attribute to the ­causes of in­equality: skills, technology, and globalization.

10 Creating a Just Society

THE ARGUMENT, RESTATED

­ umans are endowed with an innate sense of fairness and reciprocity. Though H ­shaped by history and culture, this understanding and appreciation of fairness is broadly shared across socie­ties. The only consistently approved violation of reciprocity found across cultures is a preference to give the interests of the worst off greater weight. Sympathy for the plight of t­ hose who are suffering, ­whether their circumstances are the result of bad luck or not, has evidently proven beneficial to communities throughout the world and across time, likely by strengthening solidarity. ­These innate tendencies, which provide the evolutionary and cultural foundations for a just society, w ­ ere forged in relatively egalitarian communities, according to the best efforts of scholars to mea­sure in­equality in the hunter-­gather socie­ties that characterized roughly 90 ­percent of h ­ uman history.1 Yet, the world is profoundly unequal po­liti­cally and eco­nom­ically, even in demo­cratic countries such as the United States with universal suffrage and robust civil rights. Health and income are highly unequal across racial and ethnic groups, social class, and occupation. Moreover, ­after declining during the m ­ iddle of the twentieth c­ entury, income in­equality has grown in the United States and most developed countries. ­There are many theories as to why in­equality is large and per­sis­tent and why it has recently increased in rich democracies: fundamental differences between racial and ethnic groups, globalization, trade, immigration, technological changes. On both the Left and the Right, it is assumed that ­free exchange (“markets”) naturally generates in­equality, ­either b­ ecause the fundamental source of productivity varies dramatically between individuals

274 • Chapter 10

(and groups) or ­because business o­ wners exploit workers. ­There are deep flaws with each of t­ hese theories, as I have discussed in e­ arlier chapters. First, racial diversity predicts higher income in­equality across developed countries, and it is no coincidence that the United States stands out among rich democracies as having excessive in­equality and that it relied heavi­ly on slave ­labor for much of its history. Any theory of in­equality that ignores race is fundamentally flawed. Race is essential to the analy­sis of in­equality in income and social status ­because racial prejudice provided the ideological and cultural justification for denying ­people of recent African ancestry basic liberties. Even a­ fter the end of slavery in the United States, African Americans ­were denied po­liti­cal equality in three fundamental ways: disenfranchisement, unequal access to public goods, and severe restrictions on their participation in markets. The first of ­these three was corrected with the end of Jim Crow and the civil rights movement, but the second two, while affected by the civil rights movement, persist to this day, as I demonstrated in chapters 6, 7, and 8. Black ­people are denied the equal provision of education and policing ser­vices, and they are denied participation in housing markets. The restriction on participation in housing markets applies to many non-­rich Americans regardless of race, but zoning laws emerged in such a way so as to disproportionately regulate and curtail the residency of black ­people. Given the power of neighborhoods in shaping economic opportunity, the oppressive regulation of housing markets by local factions of mostly white homeowners is a profound injustice that runs through all spheres of American economic and social life. As for changes in the income distribution, if high and rising in­equality was primarily the result of g­ reat innovations and entrepreneurial success that benefited the public broadly but allowed certain economic stars to get rich, it would arguably be of l­ ittle concern. This does not characterize the situation in the United States for several reasons. First, most individuals in the top one ­percent of income earners are not innovators: They have not in­ven­ted products or systems, and the ser­vices that they sell—­fi nancial, medical, or ­legal—­have not raised the value to cost ratio for society. Second, rising in­ equality has not caused a surge in economic growth. Rising in­equality has coincided with—­and perhaps even contributed to—­a slowdown in productivity growth. As economist Robert Gordon has shown, the last few de­cades have been characterized by weak productivity growth and lackluster innovation, relative to the 100-­year period lasting from 1870 to 1970.2 It cannot be

Creating a Just Society  •  275

true that the top one p­ ercent have only been reaping gains from their own innovations—­because the innovations that have actually materialized have been of ­little economic value. Thus, across most rich democracies, slow economic growth combined with rising in­equality has resulted in very weak income growth for ordinary ­people. In many of t­ hese countries far-­right nationalist politicians and, in some cases, far-­left populist politicians have gained traction with middle-­ income voters whose financial circumstances have not improved nearly as much as t­ hose of professional elites. Slow growth and rising in­equality are linked to the same under­lying cause: the rise of the ser­vice industries as a share of production (as opposed to manufacturing, energy, and agriculture) and the glaring inefficiencies in elite professional ser­vices, which are created by their power­ful interest groups. Across the developed world, a disproportionate percentage of the gains from economic growth are g­ oing to elite professionals, and in­equality is highly correlated with the overpayment of elite professionals. As I have shown, merit—­genes, intelligence, and work ethic—does not explain why elite professionals are d­ oing so well. Th ­ ere is an alternative explanation: the influence of the professional associations, and in finance, the influence of regulators, who defer to the industry. As the economic theorist Mancur Olson famously argued, interest groups—­such as lobbying associations—­can often win against the interests of the public ­because they have strong incentives to take “collective action,” or to shape the laws that affect their members, so they w ­ ill invest heavi­ly in d­ oing so.3 For the average citizen, any one industry or product is a very small f­actor in his or her life, accounting for but a modest fraction of expenses and absorbing even less as a m ­ atter for concern and attention. In short, we are all busy with our families, friends, work, and lives and ­can’t possibly guard against distortions to markets for this or that product or ser­vice, ­unless it is how we make a living. Thus, lobbying associations pre­sent their arguments to legislatures largely unopposed, and the arguments seem even more convincing when they are accompanied by campaign donations, as they typically are. In a recent book, Steven Teles and Brink Lindsey update and extend Olson’s argument to point out the way in which the economy is “captured” by interest groups, including home­owners’ associations that oppose open housing markets; large banks that lobby for bailouts and regulations that make it impossible for small banks to compete with them; and phar­ma­ceu­ti­cal

276 • Chapter 10

companies that lobby for patent rules that preserve their mono­poly rights for vital medicines and allow them to charge as much as insurance companies can bear to pay.4 They point to the need for countervailing forces that represent the public, as well as well-­staffed professional legislative teams and analytic organ­izations that are sophisticated enough to see through and challenge the arguments of lobbyists. I agree with ­these points and offer my own vision for what specifically should be changed to make democracies more just and egalitarian. I focus on the United States ­because it is the country I know best, but I view ­these proposals as having much broader applications.

THE NEED FOR INSTITUTIONS THAT OPERATE IN THE PUBLIC INTEREST

U.S. leaders across po­liti­cal parties have called for smarter regulations. In 2011, President Obama signed Executive Order 13563, which states: “[The regulatory system] must identify and use the best, most innovative, and least burdensome tools for achieving regulatory ends.”5 The Progressive Policy Institute, a think tank, has sensibly proposed creating a federal regulatory improvement commission to address regulatory accumulation, and a current bill in the Senate would create such a commission.6 Another proposal from the Brookings Institution would create a permanent Congressional Regulation Office, complementing the work of the Congressional Bud­get Office.7 Of course, reforming regulations does not mean eliminating them. Markets cannot function without rules. Regulations should be designed to make markets function optimally. In the early twentieth ­century, “muckraking” journalists such as Ida Tarbell inspired the creation of the Federal Trade Commission and antitrust legislation by documenting how large corporations ­were distorting markets by eliminating competition and raising prices on consumers.8 ­Today, the Federal Trade Commission and other regulatory bodies serve the public and enhance productivity by advocating for competition in healthcare and other markets.9 The capacity and influence of ­these bodies should be strengthened and their scope expanded. The much-­maligned media also needs to be strengthened, as the example of Ida Tarbell indicates. Challenges to collective business interests are likely to fail ­unless accompanied by focused public attention. Sometimes, it must be said, the interests of businesses also need to be defended against a mis­

Creating a Just Society  •  277

informed public. The internet and social media companies have weakened newspapers by eliminating much of their advertising revenue. Solutions must involve finding new ways for high-­quality news to be identified, recognized, and financially supported by a broad cross-­section of interests so that no one funding source (e.g., government, industry, or philanthropist) can exert dominance on the news. Crucially, the above institutions also need to be in place in state and local governments. Presently, a ­little bit of lobbying money—­and orga­nizational expertise—­can go a long way in influencing state legislators or city council members. It is not an exaggeration to suggest that state professional licensing laws and market-­access regulations are written largely by the most power­ ful member associations. Likewise, homeowner associations often mobilize opposition to affordable housing proj­ects or zoning reforms. For ­these reasons, states need their own institutions to seek out and challenge laws or po­liti­cal efforts that distort markets at the expense of the public.

MARKET-­A CCESS REFORMS OF SPECIFIC INDUSTRIES

­ ecause industry-­specific rules have been distorted by the shared interests of B ­those most affected, the solutions to income in­equality must also be industryspecific. The overarching theme of t­ hese reforms is to open markets so that currently prohibited mutually beneficial exchanges are allowed to take place. For workers, market access means allowing nonelite professionals to offer goods and ser­vices that they are appropriately trained to offer. Their training is appropriate if it allows them to consistently and reliably provide the ser­ vice. While some take a dim view of all licensing, especially the licensing of nonelite professions, licensing laws do afford impor­tant protections to buyers, which elevates market efficiency, and the introduction of licensing restrictions often predicts greater entry, especially for ­women and black ­people in the United States, as sociologist Beth Redbird’s analy­sis concludes, ­because it clarifies the skills, credentials, and pathway needed to gain entry.10 The challenge with such rules is to make sure they are relevant to legitimate concerns about quality and do not systematically block beneficial exchanges. For consumers, market access means being f­ ree to buy the goods or ser­vices from foreign or domestic providers that can reliably provide ­those products at the highest quality-­to-cost ratio that consumers are willing to pay for. As a result, incomes go up, b­ ecause real income is nothing more than purchasing

278 • Chapter 10

power. Opening up markets to greater foreign or domestic competition tends to accelerate business closings (and often openings), but when citizens are empowered by education and training, they readily find new work. Since eliminating foreign and domestic competition would be a disaster for the least off—­who, like every­one ­else but more so, benefit greatly from innovation and greater purchasing power—­trade protectionism is immoral from the perspective of justice. Layoffs are often damaging, but employers, ­unions, training institutions, and regional and federal policymakers can and should work cooperatively to ease transitions for workers to another job or a new field. One challenge that must be overcome is the high and rapidly growing cost for higher education. Freeing up public funding for high-­quality but low-­cost training providers—­particularly ­those offering online courses—­would be one way of overcoming this challenge. Aside from competition from foreign goods producers, competition from foreign workers brings its own set of concerns, as I argued in chapter 4. A complete deference to markets would eliminate borders entirely, allowing mi­ grants to come and go and work as they please, but this would be unjust for two reasons: First, it would expose low-­income workers to excessive competition, transferring income to the rich, and second, it would greatly weaken po­ liti­cal equality, ­because noncitizens would not have the full set of rights and entitlements as citizens, since the basis for those rights rests on the reciprocal obligations of citizenship. A solution that acknowledges both the market’s needs and its social limitations would be to greatly expand the availability of l­egal pathways for low-­income workers—­especially to Mexican and Central American populations—­while empowering t­ hose workers and other immigrant groups to gradually attain the rights of full citizenship ­after meeting reasonable criteria. Counterintuitively, the best available evidence shows that legalizing the flow of migration would result in a net decrease relative to the current situation, which deters undocumented mi­ grants from returning to their home country. Healthcare and Law

The most far-­reaching reform of the U.S. healthcare system that would lower costs and enhance its value would be the replacement of the insurance model with a group-­practitioner membership model. This would lower healthcare costs by eliminating the wasteful overhead associated with insurance payment pro­cessing and removing the incentives for physicians to overprescribe treat-

Creating a Just Society  •  279

ments and tests. Competition between groups to attract and retain patients would keep quality high. The government could pay all or some portion of the membership fee for low-­income h ­ ouse­holds or even provide something like universal membership vouchers for the entire population. A second-­best solution would be to ­either move to a more efficient single-­ payer system, which is common in other rich countries, or to implement a streamlined universal billing and payment system that would apply to government-­and privately funded insurance. ­Either one would greatly increase the efficiency of information-­technology solutions in healthcare, which are currently ­limited by the idiosyncratic complexities of each insurance provider—­complexities that themselves serve as a way for insurance providers to limit treatment and payment to providers. ­Whether group cooperatives or single-­payer insurance manages healthcare financing, eliminating arbitrary limits on hospital competition is another much-­needed reform.11 These changes—­like many innovations—­would reduce income in­equality by bolstering the purchasing power of all Americans. Innovations—­defined as increases in the quality-­to-­cost ratio of goods or services—­tend to dis­ proportionately benefit lower-­income groups b­ ecause lower-­income h ­ ouse­holds spend a larger share of their income than the affluent. Other policy changes directed at making occupational licensing rules serve the public would both lower the cost of healthcare and ­legal ser­vices and more directly redistribute money away from top-­income earners to middle-­ income workers, patients, and clients. Th ­ ese changes would reduce income in­equality in almost ­every demo­cratic country, while also raising living standards more broadly. The most urgent needs for licensing reform entail empowering licensed nonelite professionals—­such as dental hygienists and nurses—to practice fully and in­de­pen­dently, when justified by their training and expertise. This would break up the mono­poly of ser­vice provision by elite professionals. A related reform would be to allow physicians and l­ awyers to become employees of organ­izations that are owned by nonphysicians or nonlawyers. Finance

Most of the discussion of finance reforms concerns eliminating the too-­big-­ to-­fail dynamics whereby mega-­banks take too many risks with other ­people’s money, squander it, and then threaten catastrophe if not bailed out. Finance journalist Rana Foroohar makes this and other points about the negative and

280 • Chapter 10

often unintended consequences of an excessively large and concentrated financial sector.12 Likewise, the economists Anat Admati and Martin Hellwig’s elegant reform proposal—­require banks to use a larger share of their own money—­would enhance stability and protect consumers from painful financial recessions.13 ­These are surely impor­tant issues, but when it comes to reducing income in­equality and redistributing money from the top one p­ ercent to ordinary investors and retirement account holders, the most consequential reforms must deal with access to hedge, private equity, and venture capital funds. It is grossly unfair that only rich p­ eople are allowed to invest in and profit from start-­ups, initial public offerings, and acquisitions. The most beneficial policy change would create a retail market for hedge-­ fund and related products. The ­legal scholar Houman Shadab has developed a number of compelling arguments in f­ avor of this for t­ hose concerned about the consequences.14 My preferred approach would be to allow SEC-­regulated companies to offer a new, less-­regulated product—­a hedge, private equity, or venture capital fund—to any customer. Th ­ ese products would have the same disclosure and reporting requirements as mutual funds (including transparency about the general strategy, withdrawal limitations, estimated risk, and fees), but they would be f­ ree from mutual-­fund restrictions on leverage or use of derivatives, and the man­ag­ers would be allowed to charge a per­for­mance fee, which would end up being much less than 20 ­percent. A second-­best approach would be to regulate the 20 ­percent fee structure for cases when individual investors are not the ones picking the hedge fund. It is unseemly that pension funds and other institutional investors are handing over 20 ­percent of retirement profits from school teachers and blue-­collar ­union workers to billionaire hedge-­fund ­owners and advisors from Goldman Sachs, often without the consent of the investors. A consumer protection agency could directly outlaw high fees, or, better still, force hedge funds to disclose their fees publicly, so that institutional investors could at least shop around and their investors could be informed about alternative, lower-­cost arrangements.

Creating a Just Society  •  281

Intellectual Property

Patenting law is another instance where the market is wildly distorted at the expense of the public. Two reforms are needed. Patents should not be granted for “functions”; instead, they should be granted only for specific means of achieving ­things. The current practice is a massive prob­lem for software patents, and most should be nullified according to an analy­sis by the intellectual property scholar Mark Lemley.15 The presence of t­ hese functional patents creates l­egal uncertainty, invites petty but terribly expensive lawsuits, and discourages entrepreneurs, who might be threatened by patent trolls, who buy up ­these meritless patents and use them to extort payments.16 A second and more radical reform would be to stipulate that patents grant an exclusive right not to entirely own a given product but rather to own some small share of it, such as a percentage of profits sold. For example, phar­ma­ceu­ ti­cal companies now use patents to block generic drug-­makers from reproducing and selling a drug at a tiny fraction of the current price. This is a disaster for the public. U.S. taxpayers and the public typically pay over $100,000 per year for a single anticancer medi­cation for a single patient, and many of ­these medi­cations have to be taken in combination.17 ­Under a reformed patent system, the original patent holder would have no right to prevent the sale of that drug by anyone but would be entitled to collect some percentage of the generic drug-­makers’ profits off ­those sales. This would preserve much of the incentive to invest in research and development while substantially lowering prices. Antiracism

The most proven and effective cure for racism and other forms of prejudice is contact, particularly among po­liti­cal equals.18 The channels include market exchange, voluntary cooperation, and working and romantic relationships. In addition to the harm it inflicts on its victims, prejudice lowers trust, depresses collective action, weakens government quality, and distorts public policy. Eliminating artificial barriers to contact is thus of vital public importance. One of the most impor­tant artificial barriers to interracial contact is created through land use laws or zoning. Local government should be prohibited

282 • Chapter 10

from regulating the density of residential housing, with few exceptions. Investments in the renovation of old housing and the development of new housing would be more rational and efficient if zoning laws did not impede them. Rational, in this case, means building at higher density where land is more valuable and where housing is scarce relative to demand. Often, this situation is found in suburban areas with access to good schools. If density ­were permitted everywhere, it would diffuse affordability and gradually eliminate most of the economic and racial segregation that exists. Evidence for this is easy to find. Less-­regulated metropolitan areas are more integrated, U.S. metropolitan areas w ­ ere more integrated before zoning was widely ­adopted, and the United States is more segregated than Eu­rope and uses anti-­ density zoning more aggressively. Since local homeowner associations are a power­ful interest group, higher levels of government must take the lead in prying open markets from their fists. H ­ ere, I list a number of pos­si­ble remedies, which range in how much they would give in to the power of homeowners; a­ ctual bills would need to consider the legitimate cases where land use must be regulated—to protect green space, for example, or mitigate traffic congestion. 1. Pass state constitutional laws that universally ban land-­use density regulations and police loopholes that would other­wise allow local governments to block market forces. 2. Provide state tax relief—­perhaps in the form of property tax rates—to local governments that agree to open up their market by eliminating antidensity restrictions, since density growth would likely result in greater per capita government expenses. I believe state governments are best suited to make zoning policy changes through amendments to their constitutions that bolster property rights by limiting zoning to rules that explic­itly regulate harm and permit any level of housing density as the default condition. Local governments could, of course, adopt t­ hese changes, but the mayor or city council members would likely face a backlash and pos­si­ble reversal if homeowner groups ­were largely against the changes. Moreover, even if large cities manage to pass zoning deregulations, citizens are still constrained by the rules put in place by suburban governments, which are even more antidensity. The federal government could also entice states or local governments to act by providing federal grants or other

Creating a Just Society  •  283

incentives. Fi­nally, the U.S. Supreme Court could overturn a ­century of pre­ ce­dent on the validity of zoning rules, but that would require justices with much stronger re­spect for property rights than the current mix. I’m justifying zoning reform as a way to reduce racism and spark healthier civic relations, but other justifications are also impor­tant: eliminating antidensity zoning laws would make housing more affordable at the same or higher levels of quality, spurring economic growth, and it would enhance the cognitive ability and lower the crime rates of groups that live in highly segregated high-poverty communities. Information—­grounded in science—is also a cure for racism, as this book argues. First, scientific advances in genomics have reversed our understanding of the ge­ne­tic origins of IQ. Whereas throughout the twentieth c­ entury and into the early twenty-­first, IQ was thought to be mostly ge­ne­tic by leading experts, now the best evidence suggests that cognitive ability is mostly environmental for a given age cohort within a given country. The ­actual percentage varies widely across time, region, and population, but the importance of the environment becomes clearer still ­after considering that environmental effects alone account for the increase by two standard deviations in IQ from the early twentieth ­century to the early twenty-­first (the Flynn effect), and environmental conditions alone explain the fact that growing up in Western Eu­rope as opposed to India or Sub-­Saharan Africa enhances IQ by roughly one standard deviation, as we see by comparing ­children who arrive at dif­fer­ent ages. The addition of multiple cohorts across multiple countries multiplies the variation in IQ that cannot be explained by genes. Moreover, ­there is no rational basis for believing group differences are in any significant way ge­ne­tic. The social status of large ethnic groups has changed places over the last 2,000 years of ­human history, which itself rules out the possibility of long-­standing ge­ne­tic differences that are impor­tant for determining living standards. The most ambitious hereditarian estimates suggest that group differences could theoretically emerge with selective breeding over 500 years, but historical facts rule out this possibility. Eu­ro­pean Jews and East Asians demonstrated lower cognitive ability than non-­Jewish Eu­ ro­pean Americans and Northern African Americans in the early twentieth ­century and ­were less likely than both white ­people and northern black ­people to contribute scientific and product innovations to the Industrial Revolution during the period. This pattern seems to have reversed sometime in the

284 • Chapter 10

second half of the twentieth ­century for Jews and slightly ­later for East Asians, as parents from t­ hose groups became more educated. Th ­ ose dynamics rule out ­simple hereditarian explanations, but at the same time, ­there is absolutely no evidence consistent with selective breeding. If anything, ­those with higher cognitive ability ­were having the fewest ­children, especially if they w ­ ere Jewish. And yet, Jewish American and Asian American IQ levels gradually increased; the most recent large-­scale survey data show that non-­ Jewish white ­people, Asians, and Jews have equivalent cognitive scores ­after adjusting for parental education, which is higher, on average, in the latter two groups. Meanwhile, African Americans w ­ ere u­ nder the boot of Jim Crow, which lasted as late as the 1970s and lingers on in the form of unfair policing and lower-­quality schooling and neighborhood conditions. ­These shifting patterns in social status are impossible to reconcile with any cogent ge­ne­tic theory of group differences. Asian and Jewish Americans could not have significantly changed their genes relative to other groups within the span of only a few generations. Moreover, the larger global population of Asians and Jews—­meaning adults living in East Asia or Israel—­exhibits average intelligence, with the exception of the Japa­nese, who perform the same as Scandinavians and share with them a po­liti­cally and eco­nom­ically egalitarian po­liti­cal system that could be replicated around the world. East Asians, Jewish Eu­ro­pe­ans, and recent African immigrants seem to succeed academically at similar high levels in the United States and ­England, despite coming from distinct racial groups. While we can confidently rule out ge­ne­tics at the group level, the explanation for their success may partly be that mi­grants tend to exhibit cultural be­hav­iors that are conducive to success compared to ­those who do not migrate, but t­ hese practices do not require the adoption of ancient religious or ethnic practices or worldviews. Instead, the practices are quite ­simple: They include be­hav­iors such as implementing routines that are proven to boost cognitive per­for­mance of c­ hildren, such as the practice of conversing, reading books, writing, solving puzzles, and performing math exercises. Many parents did not inherit ­these cultural practices from their own parents, and in the case of illiterate slaves, serfs, or peasants, it would have been nearly impossible to gain literacy, without external help. Still, poor parents have shown remarkable willingness and ability to adopt cognitively stimulating be­hav­iors and apply them to their ­children when given even the slightest encouragement in that direction—­from ­things as ­simple as periodic text

Creating a Just Society  •  285

messages. More sustained interventions, therefore, would be more helpful still, such as regular home visits from nurses who are experts in childhood development. Recall how in the aftermath of the U.S. Civil War, ­there was an explosion in literacy and enthusiasm for learning among recently freed slaves; their ­children frequently became inventors and entrepreneurs, at least if raised in the North, and thereby made massive contributions to economic development. Education

At this heart of this book is the claim that equal access to education would have profoundly pro-­egalitarian effects on the distribution of IQ, income, and health, among other good ­things. ­Because of segregation by class and race, equal access to high-­quality education is a significant prob­lem in the United States and elsewhere. Moreover, early childhood education is typically not publicly funded, though the benefits to the public greatly exceed the costs and benefit c­ hildren with less-­educated parents disproportionately.19 A more general prob­lem with education in the United States is that the teaching profession is unattractive to top students.20 One issue is pay. Teachers are paid much less than workers with similar levels of education and cognitive ability. This is especially the case for teachers of the youn­gest students. Primary school teacher salaries are 63 ­percent of the salaries of other workers with a tertiary degree. It is barely higher for secondary school teachers. The average OECD country pays teachers closer to the “market” level for their education—78 ­percent. Top-­scoring countries like Finland pay teachers much better; t­ here, primary school teachers earn 91 ­percent of the average for tertiary workers, and upper secondary teachers are paid above the average. The same is true of ­Korea.21 Moreover, pay in the United States is closely linked to se­niority, rather than per­for­mance, making the teaching profession particularly unattractive to ­those who expect to be top performers. Using a combination of best-­practice observational evaluations, perhaps supplemented with test scores and personal judgment from supervisors, could go a long way to aligning pay incentives with merit. Low pay is not the only unattractive feature of the teaching profession in the United States. Well-­intentioned efforts to reduce racial in­equality and

286 • Chapter 10

hold schools more accountable have led to excessive testing. Instead of one test at the beginning of the year and one at the end, teachers have to administer standardized federal, state, and district exams throughout the year, some with high stakes for the school and teacher. This creates strong incentives to teach to the test. Students, meanwhile, are not actively learning while taking tests, and explicit test preparation is likely less useful than the deeper development of cognitive skills. With both lower and higher education, excessive administrative costs have crowded out funding that should be ­going to teachers, while still increasing the private and public burden of financing education. Universities now employ more highly paid professionals who do not teach than they do teachers, and the administrative staff, whose contribution to learning is at best indirect, commands the highest salaries.22 Rapid inflation in the cost of higher education over the last three de­cades is primarily the result of this shift ­toward excessive nonteaching expenses, which itself is encouraged by the indiscriminate public subsidization of higher education through federal student loan and aid programs. The public should contribute to higher education financing, ­because it benefits greatly from it, but government funding agencies need to be much more sophisticated about aligning pay with per­for­ mance and accrediting innovative lower-­cost models. Another key prob­lem in education is how to help students who are at risk to pursue their education to the fullest or develop high levels of cognitive ability. As discussed e­ arlier, the environmental experiences that lead to character and cognitive development w ­ ill never be exactly equal, nor should they be. ­Children with greater natu­ral ability or proclivity for certain fields should be given opportunities to develop ­those talents at the expense of other areas, and parents should be ­free to have authority over the educational and cultural development of their ­children. A po­liti­cally egalitarian society re­spects differences in natu­ral abilities and f­ amily circumstances and offers help to t­ hose who need it to maximize natu­ral talents from what­ever starting point. Thus, parents, particularly ­those with low levels of education, should be offered access to proven programs on how to create a healthy pregnancy and postnatal home life. Scalable low-­cost interventions such as book-­giving programs and text-­messaging reminders appropriately balance parental autonomy with the solicitude that comes from understanding the vulnerable as worthy of re­spect and support.

Creating a Just Society  •  287

The controversial educational reform efforts undertaken in the United States—­pay based on test scores, expansion of de-­unionized charter schools, and experiments with publicly funded vouchers that can be used for private schools—­should be seen in this light. Insofar as the po­liti­cal goal is to weaken ­unions or lower public expenditures on education, the reforms should be opposed. But if legislatures can come to a fair solution for how to align pay with per­for­mance for ­unionized teachers, ­there is no reason why charter schools could not be governed by the same structure. Without weakening the commitment to public universal education, vouchers, meanwhile, could be reserved for students in special circumstances, such as ­those with special education needs that could be better met in a private school or low-­income ­children who live near a high-­performing private school but are assigned to a low-­performing public school. In addition to ­these modest reforms, financial support for validated teacher and principal training or certification programs would enhance teacher quality. Respecting ge­ne­tic, cultural, or personality differences in a po­liti­cally egalitarian community, while rejecting the unfounded idea that IQ is a mea­sure of ge­ne­tic potential, has another impor­tant implication for policy: ­There is no reason to consider c­ areer and technical education as undignified. Th ­ ere are many skilled jobs that typically do not require a bachelor’s degree. Th ­ ese jobs may be as nurses, mechanics, medical or laboratory technicians, electricians, or information-­technology systems analysts or technicians.23 In addition to preserving health and maintaining the machines and infrastructure that the public relies upon, ­people in ­these roles make fundamental contributions to research and development and innovation, through their expert advice and support of academically trained professionals, as Phillip Toner has documented.24 Policymakers would do well to recognize the value of t­ hese roles and provide secondary school students a clear pathway to acquiring t­ hese skills and working in ­these fields, if they so choose. The ­Virginia Beach School District on the coast of southeastern ­Virginia provides an excellent example of what is pos­si­ble. The district allows any student in its 11 secondary schools to take ­career and technically focused classes at a state-­of-­the-­art community college. The educators are confident that the classes are directly linked to practical skills ­because the course work prepares them for certification exams created by businesses or associations. In the 2014–2015 school year, the district tested secondary students in 111 dif­fer­ent industry and professional certification

288 • Chapter 10

exams, 105 of which ­were for technical subjects related to science, engineering, or technology. Of 12,000 exams taken by just over 2,000 students, 81 ­percent ­were passed, resulting in a valuable and meaningful credential and set of skills that secondary s­ chool students can immediately use to demonstrate competency to employers or use as a foundation for further education or entrepreneurship.25 Antipoverty

I’ve argued that reciprocal ethical norms require that a just society provides every­one the opportunity to do the work that aligns with their natu­ral ability. Nonetheless, it needs to be recognized that bad luck—­including ge­ne­tic or environmental disadvantages that weaken the cognitive ability or character traits conducive to success—or bad choices sometimes lead to sustained poverty, disease, ­mental illness, or disability, even for ­those who have been born into highly advantaged circumstances. A just society protects its most vulnerable members by drawing from the resources of its strongest. This is in the long-­run interests of every­one ­because bad luck can strike e­ ither oneself or one’s descendants, and it is fair ­because it balances rewards for merit with the partial amelioration of misfortune. The “welfare states” of Eu­rope and the United States recognize t­ hese princi­ples and have built institutions in accordance with them. The challenge is to make them both effective at promoting well-­being and sustainable financially. In general, aligning ­these programs to foster decentralized mutually beneficial exchanges ­will accomplish ­those ends better than top-­down strategies that create complicated rules about who gets benefits, who provides them, and how they can be used. For example, requiring housing developers to build affordable housing for the poor is a less effective and financially sustainable strategy than a program that gives poor ­people a housing voucher to support their rent (or lower their mortgage payment) wherever they choose to live. It is unfair to force private businesses to shoulder the burden of public welfare beyond their mandated tax payments, and it is less effective, ­because it gives them strong incentives ­either to avoid certain proj­ects that would benefit ­others or to overcharge other customers to make up for the revenue loss. Similar logic could apply to employer requirements to pay healthcare benefits.

Creating a Just Society  •  289

Another aspect of antipoverty policy relates to adult training. Egalitarian countries such as Denmark have been successful in combining unemployment insurance with requirements to undergo publicly funded training in pursuit of a new line of work for p­ eople who can no longer find a job in their chosen field. This demonstrates a strong reciprocal norm: Able-­bodied beneficiaries of public generosity are expected to give something back by investing in skills that w ­ ill be useful to society. Accordingly, ­because common ­people are valued and not seen as genet­ically inferior, the public has confidence that investing in their retraining ­will be worthwhile.

MARKET EGALITARIANISM

I define a just society as one that grants its members equal opportunity to realize and apply their talents u­ nder conditions of po­liti­cal equality, defined by basic civil liberties and access to markets. I maintain that this background structure leads to a relatively egalitarian distribution of income and health. The set of ideas ­behind this could be thought of as market egalitarianism, ­because of the dual emphasis on equality of rules and the reliance on mutually beneficial market exchanges. ­People ­today are rightly skeptical about claims about justice. On the Left, ­there is suspicion of power; in recent history, power­ful white-­dominated nations have imposed their w ­ ill on o­ thers—­via slavery and colonialism—at the expense of justice, even while claiming to represent it. On the Right, ­there seems to be an exhaustion and frustration with claims that racism is ­behind ­every unfavorable outcome for black p­ eople or other nonwhite groups. The phrase “social justice warrior” has become a derogatory term to describe ­people on the Left, who, from the perspective of conservatives, engage in self-­righteous, illogical, and emotionally laden public discourse. I have a very dif­fer­ent starting point. Market egalitarianism is the foundation, and a just end is what naturally follows from it. This starting point aligns most closely with experiences that s­ haped our DNA and our moral cooperative culture and institutions. From ­there, reciprocity ­toward community members and even strangers follows naturally, as does generosity ­toward the weak. As I’ve argued throughout, market egalitarianism does not set out to guarantee equal economic or social outcomes, but it would make society far more

290 • Chapter 10

egalitarian eco­nom­ically and thus accomplish many of the goals of the socialist Left. The reason is that when decentralized mutually beneficial exchanges are allowed to govern markets, ­people tend to be rewarded on merit rather than po­liti­cal power, and merit is distributed equitably when given a chance to develop. This distribution is not perfectly equal, of course, but equal enough such that every­one except the most profoundly disabled can find jobs that are broadly productive and socially useful. The most productive and clever among us—­even the luckiest—­should be rewarded when their inventions and ventures increase the ratio of product quality to cost, but ­doing so w ­ ill never lead to the massive in­equality observed ­today in the United States, Latin Amer­ i­ca, and much of the world. Extremes in in­equality arise when pay is systematically divorced from merit. Even in highly egalitarian socie­ties—­such as Scandinavia and Japan—­ entrepreneurs, executives, and elite performers have opportunities to distinguish themselves and make much higher incomes than ordinary professionals. The difference between t­ hese countries and the United States is that t­ here are far fewer ­people who reach that level as a result of antimarket distortions, so overall in­equality remains much lower. As for more conservative goals, market egalitarianism provides a defense of liberty, private property, and mutually beneficial exchanges, as well as a protection against excesses and abuses of government power. From a cultural perspective, market egalitarianism creates conditions that allow p­ eople to solve their prob­lems and meet their needs as they see fit. For many, that ­will involve the continued embrace of religious and cultural traditions that guide their values. At the same time, the integration of p­ eople, culture, and information that comes from decentralized housing market choices, from migration, from trade, and from mundane exchanges within cities tends t­ oward harmonious cooperative interactions, trust, and mutual re­spect. Insofar as the religious and cultural traditions valued by conservative Eu­ro­pe­ans promote healthy and appealing lifestyles, they are likely to be embraced by immigrants and ­those from other cultures. At the same time, the native Christian populations could benefit by adopting some of the parenting practices of Eu­ro­pean Jews, Asians, or Sub-­Saharan Africans with regard to actively promoting literacy and numeracy in their ­children. A diverse society allows groups to learn from each other and allows good ideas—­whether ancient or modern—to emerge from cultural competition.

Creating a Just Society  •  291

One cultural tradition that deserves to be rejected is the doctrine of racial superiority, which is closely linked to hereditarian theory. It has haunted Eu­ ro­pean and American society like an intellectual plague for several centuries now and spread outside it to Asia and elsewhere. It has caused tremendous harm by justifying po­liti­cal in­equality, even genocide; it has created and sustained an underclass of citizens, squandering their natu­ral ability; it has deprived millions of would-be immigrants of the opportunity to migrate; and it has drastically eroded trust and cooperation not only between Eu­ro­pean white p­ eople and nonwhite ­people and Eu­ro­pean Jews and non-­Jews, but even among them. Po­liti­cal dysfunction in the United States and much of Eu­rope is closely tied to differences in opinions about how to relate to minorities and the foreign born, and hereditarianism exacerbates ­those differences. Less seriously, the doctrine is to blame for the stifling awkwardness and fear that permeate college campuses with re­spect to open discussions of group social status differences and how to mitigate them. The painful implications of white supremacy as a doctrine makes it poisonous to take in and dwell upon at length. The prob­lem then is that sophisticated theoretical claims made by hereditarian social scientists and intellectuals are never engaged openly and appear edgy and seductive to many who come across them on the internet unaware of their weaknesses. As I’ve argued, ­there is no logical or scientific basis for hereditarianism. The best available ge­ne­tic research shows that the heritability of IQ has been greatly exaggerated, even within the same age group and by greater amounts across generations or socie­ties at dif­f er­ent levels of economic development. The environmental origins of the massive two-­standard-deviation increase in cognitive ability over the past 100 years are themselves power­ful evidence that genes cannot account for more than a modest share of variation across individuals. Most importantly, t­ here is no logical basis for believing that group differences are ge­ne­tic in nature, in any impor­tant way. The instability in group social status and intellectual achievements over the last 100 years disproves it; the consistently low and then high per­for­mance of American Jews and Asians disproves it; the intellectual achievements of African Americans ­after the Civil War but before Jim Crow disprove it; the current intellectual achievements of African immigrants in ­England and the United States disprove it; and the clearly demonstrated, replicable, and undeniable gains in IQ

292 • Chapter 10

that come from exposure to high-­quality education and educated parenting disprove it. Cognitive ability, like e­ very impor­tant ­human trait, has a ge­ne­tic basis, but exercise, practice, and training are most impor­tant in determining the expression of fitness, and most ­people can become fit—­fit enough to perform any occupation. Moreover, noncognitive character traits such as conscientiousness, emotional stability, enthusiasm, and integrity are at least as impor­tant as IQ in preparing ­people for success in life, mea­sured through income or health, and the evidence presented h ­ ere shows that t­ hese traits are distributed evenly across race, ethnicity, and gender, despite wide variation in the parenting be­ hav­iors that foster IQ development. If p­ eople are ­going to reach their optimal potential, we would do well to foster t­ hese noncognitive skills in t­ hose at risk of not fully developing them. We know they are fostered by a loving upbringing, and it seems likely that they are enriched by the best artistic and religious ele­ments of culture: ­great books, films, plays, works of art, sermons, and religious teachings. The discipline and grounded moral tradition of military life may also boost noncognitive skills. Quite likely school-­based practices and therapeutic interventions can help remediate deficits, as the scholarly lit­er­a­ture indicates.26 This is an area in ­great need of additional research, as the focus of education has been almost exclusively on cognitive skills. In a promising development school districts in California have begun mea­sur­ing social-­emotional skills—­such as self-­management—­alongside traditional cognitive per­for­mance as a first step in understanding how to effectively develop them.27 The malleability of cognitive ability, the even spread of psychological traits conducive to realizing one’s capacity, and the responsiveness of ­people to educational opportunities all suggest that the vast gaps in social and economic power that exist in modern democracies are not inevitable. Nor are they desirable. The large gaps have come about primarily through the enactment of deeply harmful barriers to ­free exchange, put in place by power­ful interest groups. The nineteenth ­century saw a bourgeois revolution against the hereditary elite. The professional class won that revolution, and billions of p­ eople have benefited greatly as a result. Yet, the professional elites have become too entrenched and are the new landed gentry, claiming entitlement to special privileges such as the right to work without being an employee and the right to

Creating a Just Society  •  293

own a mono­poly on ser­vices o­ thers could do quite well. Quite likely, a feeling of natu­ral superiority set in, given that many of the professional associations gained prominence at the same time that racial science gained traction. The nonelite professions—­those not afforded mono­poly privileges, including engineers, computer programmers, midlevel man­ag­ers, technicians, nurses, and educators, as well as vari­ous para-­professionals—­need to take part in the next stage of revolution. They, like the working class, would stand to benefit by eliminating the privileges that siphon away their spending power to elite professionals, restoring a common understanding that every­one has something to contribute, and creating the po­liti­cal equality promised by republican ideals.

APPENDIX

This appendix is written for scholars, students, journalists, and anyone ­else interested in the technical details of the data used in the book. Readers should be aware that much of the data and related replication code (written for STATA software) are available freely for download at https://­ github​.­com​/­jtrothwell​/­Republic​_ ­Equals​/­, and the source and description of each variable accompany ­those files. The files (or supplemental materials, as I ­will refer to them below) include topics that are not addressed in what follows, such as my analy­sis of National Longitudinal Survey of Youth (personality, IQ, and their relationship to health, income, race, and ethnicity) and pre-­kindergarten schooling and test score data. The text below discusses the following topics: 1. Analyzing who is in the top one p­ ercent in the United States and around the world pre­sents methodological challenges. Th ­ ese are explained, along with why I think my analy­sis is valid. 2. For p­ eople skeptical that in­equality is a meaningful summary mea­sure or curious about the relationship between dif­fer­ent mea­sures of in­ equality (top 1 ­percent shares versus Gini coefficient), I provide a brief discussion and analy­sis that aims to show that in­equality can be ­mea­sured reliably across methods and offers valid insight into living conditions. 3. The section “In­equality, Slow Growth, and Falling National Confidence” pre­sents results from a s­ imple regression analy­sis. The full results are available in the supplemental materials and are consistent with my argument that slow growth and in­equality predict low levels of confidence in national government.

296 • Appendix

4. The section “What Explains In­equality?” sketches the main theoretical argument of the book and pre­sents empirical results from a historical accounting exercise of which occupations and industries explain the rise in the U.S. one p­ ercent. It then turns to an international analy­sis and discusses correlations and regression results that look at the most-robust predictors of in­equality across OECD countries. ­These findings suggest that elite professional power and racial diversity predict higher in­equality; meanwhile, government spending (as a share of GDP) and innovation (mea­sured by patenting rates) predict lower in­equality. This evidence is consistent with the arguments made in the main text. Mea­sures of globalization, the power of business ­owners as opposed to workers, and gaps in cognitive ability are unrelated to in­equality, which is consistent with my fundamental argument: Markets do not inherently lead to mass in­ equality; mass in­equality has been created through racial oppression and isolation—­blocking minorities from public ser­vices and markets—­and the disruption and control of certain markets by elite professionals.

IDENTIFYING WHO IS IN THE TOP ONE P ­ ERCENT

For levels and trends in top-­one-­percent income shares and for analy­sis of thresholds, I rely on the World In­equality Database (WID), which is assembled by Thomas Piketty and his collaborators using national tax data.1 Unfortunately, tax rec­ords do not usually rec­ord information about the occupation or industry of ­those who file their taxes, and t­hese data are not available in the WID. Thus, to analyze top earnings by occupation and industry, I relied on government surveys. The American Community Survey (ACS), with the data packaged and managed by IPUMS USA, was my primary data source for analyzing the U.S. one ­percent.2 For international analy­sis, I relied primarily on the Luxembourg Income Study (LIS).3 ­There are several issues that arise with t­ hese sources, which warrant comment and clarification.

Appendix • 297

Issue 1: Reporting Thresholds for Income Sources

One limitation of the American Community Survey is that under­lying income sources (wage/salary, investment, and business being the most impor­ tant for the affluent) are top-­coded at the 99.5th percentile (the top 0.5 ­percent) for each state to make it difficult to identify individuals. For the analy­sis of the one p­ ercent in this book, I have concluded that it is not a significant prob­ lem for several reasons. First, total income is not top-­coded, only the under­lying sources, so the only way someone who is ­really in the top one ­percent can be misclassified is if he or she happens to have top-­coded business income or wage income in a state where the top codes fall below the national one-­percent threshold (of $315,700) but still does not make it to the top one p­ ercent, once all other income sources are included (­table A.1). ­There are only four states for which the wage/salary income top code for 2016 falls below the national top-­income threshold of $316,000 for the entire period of 2012–2016 (West V ­ irginia, Alaska, Mississippi, and Vermont), but almost every­one with top-­coded wage income makes it into the one ­percent when their salary income is combined with another income source. The business income threshold falls below $316,000 for 36 states in 2016, but again, ­there are very few ­people who have top-­coded business income and do not reach the one ­percent. To be more precise, only 216 respondents out of 12 million with income data are potentially misclassified as not being in the one p­ ercent despite having top-­coded salaries in the 2012–2016 census database. Another 1,789 are potentially misclassified ­because they have top-­coded business income data, but do not reach the one ­percent. This is 0.01 ­percent of U.S. residents with income. A higher number (7,447) have top-­coded investment income, but are not in the one ­percent. They appear to be mostly retirees living off their investment income, as 81 ­percent are out of the ­labor force and 62 ­percent are at least 65 years old. This leaves just 1,139 workers who, based on investment income alone, could be misclassified as not being in the top one ­percent. Second, if you believe that the one-­percent threshold is best established through fiscal income (defined as the income p­ eople must report to tax authorities) mea­sured from the World In­equality Database, then identifying the top one p­ ercent of fiscal income holders from the American Community Survey is even more straightforward ­because the threshold is lower. In the World

298 • Appendix

­ able A.1.  Income thresholds to qualify as a top-one-percent earner in T the United States using World In­equality Database and Census in 2014 American Community Survey Wage and salary income Total income

$221,616 $315,700

World In­equality Database Capital income reported on tax returns per adult on return ­Labor (or employee) income reported in tax returns per adult on return Total income reported on tax returns per adult on return Total national income (including non­taxable l­ abor and capital income) per adult on return

$242,775 $264,286 $283,749 $469,256

Source: World In­equality Database, accessed via WID STATA program on May 10, 2018. The population is adults aged 20 and over. Steven Ruggles, Katie Genadek, Ronald Goeken, Josiah Grover, and Matthew Sobek. Integrated Public Use Microdata Series: Version 7.0, 2012–2016 American Community Survey. Minneapolis: University of Minnesota, 2017. https://­doi​.­org​/­10​.­18128​/­D010​.­V 7​.­0. In the 2014 American Community Survey, e­ very state’s top-­code threshold for l­ abor income is above the l­ abor-income threshold for the one ­percent except West ­Virginia. Additionally Montana and New Mexico’s thresholds fall below the total-income threshold in the WID. For both, almost any amount of income from business owner­ship or dividends would be enough to push a high wage earner into the top 1%.

In­equality Database, the U.S. national one-­percent threshold per adult h ­ ouse member for taxable (fiscal) income is $284,000 for 2014, which is below e­ very 2016 wage and salary top code in the census and greater than only three states for 2012 top codes. The census threshold is likely somewhat higher ­because the tax rec­ords are often not mea­sured for individuals, but rather for tax returns, which often have multiple adults (e.g., a husband and wife). This is akin to mea­sur­ing ­house­hold income per capita, which w ­ ill tend to be lower than individual income for top earners, since the other adults with whom they share a h ­ ouse­hold ­will typically earn less (­table A.1). I conclude that census reporting thresholds are not a serious obstacle to accurate analy­sis of who is in the one ­percent. With 119,130 individuals over the age of 20 identified as being in the top one ­percent, ­those who are potentially misclassified amount to a tiny fraction of the total. Issue 2: Missing Capital Income in the Census Data

Another limitation of the American Community Survey is that it does not include income data from the sale or exchange of a “capital asset,” such as a stock or owner­ship stake in a com­pany. It does, however, include data from

Appendix • 299

business income, including dividends, royalty payments, and rental payments. In effect, this means that the ACS includes only income that flows to individuals as e­ ither o­ wners or workers, but does not include income from sales of property u­ nless directly tied to a business controlled by the individual. It also does not include employer or government allocations of funds to an individual’s retirement account, the implicit value of owning a home (“owner-­ occupied rent”), the distribution of taxes on business, or the value of compensation through public or private health insurance plans. Recent work from Thomas Piketty, Emanuel Saez, and Gabriel Zucman includes t­ hese concepts as part of income (they call it “national income”) and reports that the U.S. top-­one-­percent threshold for individual pretax income is $470,000 for 2014.4 The threshold greatly exceeds taxable income b­ ecause it includes all of t­ hese other concepts, which, in most cases, are not like income that one can use to buy products, at least in the pre­sent. The impor­tant point is that the sale of assets could be an impor­tant source of income for rich ­people and be systematically allocated to certain occupations, such as executives, in ways that bias my ­earlier results. To better understand how dif­fer­ent income concepts and sources relate, I combine IRS zip-­code data—­which is available to the public—­with census data. I use adjusted gross income, which corresponds with what Piketty and co-­authors describe as fiscal income. At the zip-­code level, mean census income has a very high correlation of 0.87 with mean IRS adjusted gross income—­and a correlation of 0.91 between the share of residents in the top one ­percent (incomes above $284,000) and mean census income. IRS gross income is also highly correlated with median home values (0.71). In addition, IRS data distinguish income from capital gains and business partnerships. Average capital-­gains income at the zip-­code level has a correlation of 0.58 with mean census income, while income from partnerships has a correlation of 0.72 with mean income (­table A.2). Th ­ ese correlations use zip-­code population as a weight, but the results are similar when giving equal weight to all zip codes. To further explore the relationship between census income at the individual level and fiscal income, I match IRS data by zip codes to the census-­ relevant Zip Code Tabulation Areas (ZCTAs) using the Uniform Data Systems (UDS) zip mapper.5 The next step is to match ZCTAs to a unit available in the census microdata. The smallest such unit of analy­sis is a Public Use Microdata Area (PUMA), which has 100,000 p­ eople. I do this using an application published by Missouri Census Data Center.6

1.00 0.64 0.26 0.26

1.00 0.30 0.43 0.52

0.51

Total income, census

0.33

1.00 0.33

Total ­family income, census

0.96

1.00 1.00

­ ercent of residents P in zip code with IRS adjusted gross income of $284,000 or higher

Sources: Correlations based on 1,176,360 individual observations. U.S. Internal Revenue Ser­vice, Individual Income Tax Statistics, 2015; Steven Ruggles, Katie Genadek, Ronald Goeken, Josiah Grover, and Matthew Sobek. Integrated Public Use Microdata Series: Version 7.0, 2012–2016 American Community Survey. Minneapolis: University of Minnesota, 2017. https://­doi​.­org​/­10​.­18128​/­D010​.­V 7​.­0.

Home value Total income, census Total ­family income, census Average adjusted gross income from IRS in zip code ­Percent of residents in zip code with IRS adjusted gross income of $284,000 or higher

Home value

Average adjusted gross income from IRS in zip code

­Table A.2.  Correlation between census income and other income concepts, 2015

Appendix • 301

The results show a high correlation between home values at the ­house­hold level and IRS adjusted-­gross-­income levels at the PUMA level. IRS income is also highly correlated with f­ amily income (.64). The correlation between IRS income and individual income is lower, but this is not surprising, ­because variation within PUMAs and within h ­ ouse­holds creates noise that lowers ­these correlations. The high correlation between IRS zip-­code income data and home values (0.52) implies that home values may be an impor­tant proxy mea­sure for the income sources not captured by the American Community Survey (e.g., capital income). Thus, as a robustness check, I report how occupations vary by home value. At the detailed level, the census data are grouped into 480 occupational categories by IPUMS USA (using “occsoc”). The correlation between the percentage of workers in each occupation who are in the top one ­percent of income and the percentage who live in a home valued in the top one ­percent is 0.73. It is likewise high (0.65) between the percentage living in a top-­one-­percent home and a top-­one-­percent zip ­code. Consistent with my e­ arlier analy­sis, physicians and surgeons, dentists, ­lawyers, and securities traders are heavi­ly overrepresented among top-­one-­ percent homeowners. While chief executives and other managerial groups are also heavi­ly represented, the additional consideration of home value does not appear to substantially change the analy­sis. In other words, while it would be desirable to explic­itly include all forms of capital income in the analy­sis of which occupations are most overrepresented among top earners, d­ oing so prob­ably would have ­little effect on the ­actual composition of top one-percent earners. Doctors, dentists, ­lawyers, and finance-­related workers are at or near the top any which way you mea­sure income (­table A.3). International Estimates

The same two issues with U.S. census data apply to census data from other countries, which I access through the Luxembourg Income Study, but I conclude they are not serious obstacles to accurate analy­sis. First, maximum income reported in each country in the LIS exceeds national top-­one-­percent thresholds in all 12 cases that I could compare against WID data using LIS income data from 2010 or ­later. Thus, reporting restrictions are not an a­ ctual prob­lem for describing who is in the one ­percent outside the United States using t­ hese data. For LIS and U.S. Census

6.4% 6.0% 5.8% 5.6% 5.0% 5.0% 4.8% 4.1% 4.0% 3.8% 3.6% 3.4% 3.0% 2.8% 2.8% 2.8% 2.7% 2.6% 2.5% 2.4%

13.3% 12.9% 11.8% 12.8% 14.2% 11.8% 13.8% 11.4% 12.8% 12.9% 12.6% 9.6% 14.4% 9.5% 13.0% 10.3% 12.1% 12.5% 8.2% 13.7%

Top 1%—­zip code, IRS income 30.3% 16.4% 1.9% 10.6% 6.9% 18.4% 13.7% 2.6% 2.6% 11.6% 2.1% 3.2% 6.0% 5.6% 4.5% 3.3% 2.7% 1.7% 1.4% 2.4%

Top 1%—­individual income, census

Sources: Correlations based on 1,176,360 individual observations. U.S. Internal Revenue Ser­vice, Individual Income Tax Statistics, 2015; Steven Ruggles, Katie Genadek, Ronald Goeken, Josiah Grover, and Matthew Sobek. Integrated Public Use Microdata Series: Version 7.0, 2012–2016 American Community Survey. Minneapolis: University of Minnesota, 2017. https://­doi​.­org​/­10​.­18128​/­D010​.­V 7​.­0.

Physicians and surgeons Chief executives Actors Securities, commodities, and financial ser­vices sales agents Financial analysts Dentists ­Lawyers, judges, magistrates, and other judicial workers Agents and business man­ag­ers of artists, performers, and athletes Producers and directors Personal financial advisors Writers and authors Property, real estate, and community association man­ag­ers Economists Financial specialists, all other Management analysts Real estate brokers and sales agents Public relations specialists Psychologists Lodging man­ag­ers Market research analysts and marketing specialists

Top 1%— home value

­Table A.3.  Occupations ranked by share in top one p ­ ercent of home values, 2012–2016

Appendix • 303

Bureau sources, the data do not allow for accurate calculations of the share of national income that goes to the one p­ ercent, since it would cut off a large chunk of money flowing to p­ eople whose income exceeds the reporting thresholds, but that limitation does not affect the description of who is in the one p­ ercent, which is the focus of this book. As for the capital gains issue, nonlabor income is roughly the same share of national income in the United States as in other OECD countries, according to data from the Penn World ­Table.7 It is unlikely that including capital income would substantially affect the occupational rankings, since efforts to do so made ­little difference in the United States. One reason that capital versus ­labor income distinctions are unlikely to affect the occupational distribution is that with the exception of t­ hose who inherit large fortunes, capital income amounts to the profits of invested savings from l­ abor income. In this sense, capital income is, in large part, earnings from past l­ abor income. Analy­ sis from Robert Carroll of the Tax Foundation finds that capital income is also far more transitory than ­labor income at the individual level. Roughly, half of taxpayers with over $1 million in taxable income reach millionaire status only once in a nine-­year period.8 ­Labor income, however, is more stable, which makes comparisons that exclude capital gains—as in the census—­ relatively attractive when describing the characteristics of top earners over time.

THE VALIDITY OF INCOME IN­E QUALITY MEA­S URES

­ hether mea­sured by the top-­one-­percent share of taxable income using tax W rec­ords or by the Gini coefficient using h ­ ouse­hold survey data, countries that are more unequal on one mea­sure tend to be more unequal on o­ thers. This is true for 26 countries with data from the United Nations World Income In­equality Database (UN WIID), LIS, and WID. The survey-­based UN and LIS mea­sures are highly correlated (0.91), and both are highly correlated with top-­one-­percent income shares (0.74 and 0.79, respectively). Within the 16 countries in the OECD with data from all three sources, the correlations are similar (0.65 and 0.74) between top-­one-­percent shares and UN and LIS Gini coefficients, respectively. In this sense, the mea­sures used in this book are reliable across data sources and methods.

304 • Appendix

In­equality mea­sures are also clearly associated with bad outcomes, such as low health and low subjective well-­being. In this sense, they are valid. For example, the Gallup World Poll asks several questions related to financial well-­being, including the following: • Are you satisfied or dissatisfied with your standard of living, all the ­things you can buy and do? • Which one of t­ hese phrases comes closest to your own feelings about your ­house­hold’s income ­these days: living comfortably on pre­sent income, getting by on pre­sent income, finding it difficult on pre­sent income, or finding it very difficult on pre­sent income? Both of t­ hese questions are highly correlated with in­equality mea­sured across all three data sources. The most unequal countries in the world—­using the UN database—­are in Latin Amer­i­ca, followed by Sub-­Saharan Africa. In ­these regions, p­ eople tend to be dissatisfied with their living standard and many find living difficult (­table A.4). By contrast, in egalitarian northern and western Eu­rope, p­ eople are highly satisfied with their living standard, and few find living difficult. The global correlation (for 101 countries) between income in­equality and the p­ ercent satisfied with their living standard is −0.27. The correlation between income in­equality and difficulty living on income is 0.42. Among OECD countries, Norway, Sweden, Denmark, and Japan have the lowest percentage of ­people who report difficulty living comfortably on their ­house­hold income, ranging from 10.3 ­percent in Japan to 6 ­percent in Norway. Meanwhile, in countries with higher income in­equality, such as Canada, the United Kingdom, and the United States, 15 ­percent, 19.1 ­percent, and 19.7 percent, respectively, report difficulty. The OECD correlation between ­income in­equality and the p­ ercent satisfied with their living standard for 30 countries is −0.37. The correlation between income in­equality and difficulty living on income is 0.33. Th ­ ere is also a statistically significant correlation between in­equality and healthy life expectancy (–.22) and life evaluation (–.34) within the OECD.9 The poor per­for­mance of the United States on subjective financial well-­ being is particularly surprising considering that the United States is richer on a per c­ apita basis than all but five OECD countries but ranks 17th on this mea­sure of subjective financial well-­being. The latest World Bank data from

Appendix • 305

­Table A.4.  In­equality and subjective financial well-­being by world region

Western Eu­rope Northern Eu­rope Eastern Eu­rope Southern Eu­rope Australia and New Zealand Northern Amer­i­ca Western Asia Central Asia Southern Asia Northern Africa Eastern Asia South­east Asia Sub-­Saharan Africa Latin Amer­i­ca and the Ca­rib­bean

Income in­equality, UN database, standardized to mean zero

­ ercent P satisfied with standard of living

­ ercent P finding living difficult with ­house­hold income

−1.11 −0.97 −0.83 −0.77 −0.45 −0.39 −0.38 −0.31 −0.28 −0.18 −0.08 0.26 0.63 1.16

0.87 0.74 0.53 0.56 0.87 0.81 0.62 0.78 0.67 0.67 0.73 0.77 0.41 0.68

0.17 0.20 0.37 0.37 0.13 0.17 0.41 0.22 0.42 0.41 0.25 0.30 0.64 0.41

Source: Data from the Gallup World Poll (columns 2 and 3) are for years 2013–2017, with sample sizes of roughly 1,000 per country per year. UN data (column 1) average the ten most recent observations up to 2017 and are from the World Income In­equality Database, Version 3.4.

2016 show that only Luxembourg, Singapore, Ireland, Switzerland, and Norway have higher incomes per capita than the United States, a­ fter adjusting for purchasing power.10 High income in­equality, presumably, is the chief explanation for why such a high percentage of U.S. residents do not feel comfortable with their living standard, despite living in such a rich country.

IN­E QUALITY, SLOW GROWTH, AND FALLING NATIONAL CONFIDENCE

To test w ­ hether growth and in­equality predict po­liti­cal dissatisfaction, I regress the change in national government confidence on income in­equality and per c­ apita income growth from 2008 and 2017. Th ­ ere is a strong negative and significant relationship between in­equality and rising confidence (t-­stat is −3.2), and a strong positive relationship between rising confidence

306 • Appendix

and higher growth (t-­stat is 3.3). The r-­squared is 0.34 for this model, which uses data from 35 OECD member countries.

WHAT EXPLAINS IN­E QUALITY?

A Theoretical Sketch

To provide a somewhat more formal theoretical framework for my argument, I refer readers to a summary of growth theory from Stanford University economist Charles Jones, as well as a working paper he has developed with Chang-­Tai Hsieh, Erik Hurst, and Peter Klenow.11 Drawing on their analy­ sis, I ­will lay out some of the logic of my own theory. Modern economic growth theory links long-­r un growth in living standards to productivity growth, which describes changes in how efficiently each ­factor of production (usually l­abor and capital) is being utilized. Productivity is often represented as ­either knowledge or technology, but Jones describes a new branch of growth theory that allows for the misallocation of resources (such as p­ eople) to govern productivity growth. It is straightforward to represent this idea in the mathematical language and context of macro­ economic growth lit­er­a­ture. For my purposes, the misallocation of talent can result from inequitable po­liti­cal power, and this, in turn, w ­ ill result in both slower productivity growth and high-­income in­equality, as p­ eople work in the wrong ­careers. To see how the misallocation of talent may arise in ways that correspond to the broad themes of this book, imagine that ­people choose their ­career in early adulthood in such a way as to maximize their lifetime health and income. To the best of their knowledge, income and health are determined by how productively they can perform the tasks of a given occupation in the economy. When choosing a c­ areer, they thus aim to pick the set of tasks that ­will allow them to work most productively, given what they know about their preferences and talent. Talent at young adulthood, meanwhile, is determined by ge­ne­tics, private environmental circumstances (such as neighborhood and familial influences), and public influences, such as environmental security, law enforcement, and education. In a po­liti­cally egalitarian society, p­ eople choose ­careers that line up closely with their talent profile, given ­these ­factors, but in a po­liti­cally unequal society, power can have a large influence in several

Appendix • 307

ways. Power is determined by group membership, which is determined by ascribed characteristics (e.g., race, ethnicity, and gender), as well as malleable characteristics (e.g., occupational or industry membership). When ascribed characteristics affect power, the accumulation of talent is distorted through the nonge­ne­tic channels mentioned (access to optimal private and public environmental circumstances), and this has a direct effect on c­ areer choice by lowering talent from what it would other­wise be. Another channel by which inequitable power distorts ­career choice is by taxing away the income of members of low-­power groups. To elaborate, prejudice against religious or racial minorities, w ­ omen, and workers in nonelite professions has resulted in lower earnings and fewer entrepreneurial possibilities than would be the case with a more equitable distribution of power. ­These income losses are predictable and deter low-­power groups from performing their most productive tasks. For members of power­ful groups, the obverse of this dynamic dominates, and they are encouraged to over-­pursue high-­paying ­careers, even when other ­careers may have been a better use of their talents, b­ ecause they receive an earnings benefit in ­doing so. Thus, ­people who would have been excellent plumbers, machinists, teachers, or scientists become doctors, ­lawyers, dentists, and hedge-­fund man­ag­ers instead. Members of power­ful groups may enjoy somewhat higher talent insofar as they can better leverage private and public resources, but they also enjoy over-­compensation for their talents as a result of access to privileged industrial or occupational groups. In this way, po­liti­cal in­equality creates income in­equality that ­favors the members of po­liti­cally power­ful groups and results in slower economic growth than would other­wise be the case: Too many p­ eople in the low-­power group are working in low-­productivity jobs that fail to utilize their talent, and too many p­ eople in the high-­power group are working in jobs that have high productivity but low macroeconomic efficiency (e.g., ­lawyers and plastic surgeons instead of engineers and software developers).12 To extend this to my theory of politics, right-­wing pop­u­lism w ­ ill become more attractive in this kind of society b­ ecause t­ hose who share partial membership advantages (say, middle-­class white males) but not the full advantages of industrial or occupational status ­will develop deep distrust of elites and be more willing to embrace xenophobia, especially if they come to believe that increasing numbers of members of the elite are from foreign groups who do not traditionally hold power. The chief po­liti­cal opponents for this group are the subset of elites who are minority members of occupational and industrial

308 • Appendix

groups. Left-­wing pop­u­lism w ­ ill also become more attractive and its chief po­ liti­cal opponents ­will be white male members of privileged occupational and industrial groups. In both cases, however, the shared distrust of elites w ­ ill make it hard for populists to distinguish between successful individuals who have earned their status through talent and ­those who have merely cashed in on their power. This makes a large segment of the voting population open to spurious policy analy­sis that links in­equality and slow growth to salient trends (such as globalization, immigration, and technology) even when they are unrelated from a causal perspective. Empirical Analy­sis of the Po­liti­cal Economy of In­equality

One way to study the f­ actors that relate to rising income in­equality is to see how the composition of the one ­percent has changed over time by sector (the products firms produce) and occupation (the type of work ­people do for their firm). I focus on 1980 as the starting point b­ ecause that is when top-­one-­­percent income shares began to rise ­a fter a long-­term fall, as documented in the World Income Database and by Census Bureau mea­sures of in­equality. The one ­percent have increasingly consisted of financial-­sector man­ag­ers, professional and other business ser­vice man­ag­ers, healthcare professionals and man­ag­ers, and ­legal ser­vice providers. If you add up the total increase in workers from 1980 to 2015 in ­these categories you get 98 ­percent of the total increase in top-­one-­percent earners (­table A.5). Keep in mind that other sectors and occupations also contributed statistically to the rise of the one p­ ercent b­ ecause some sectors had a negative effect. Large reductions in the number of top income earners ­were found among craft and construction workers, farmers, and retail and ­wholesale man­ag­ers. Overall, this provides more evidence that a fairly narrow group of domestically focused professionals and man­ag­ers—­particularly man­ag­ers in the financial and healthcare sectors—­explains the rise of the one p­ ercent in the United States. Any coherent theory of income in­equality must focus on why elite healthcare professionals, financial-­sector man­ag­ers and professionals, and ­lawyers are so overrepresented now and account for such a large share of the increase. This makes the globalization story look rather weak.

­ able A.5.  Accounting for change in number of workers in top one T percent of U.S. income earners by high-level occupation and sector from 1980 to 2015 Broad occupational category Managerial Managerial Professional specialists Managerial Professional specialists Managerial Professional specialists Managerial Professional specialists Professional specialists Finance and sales Professional specialists Professional specialists Professional specialists Professional specialists Finance and sales Managerial Administrative Managerial Finance and sales

Broad sector Financial ser ­vices Other professional and social ser­vices (e.g., engineering, accounting) Healthcare ser­vices Business and repair ser­vices (e.g., advertising, computer ser­vices) ­Legal ser­vices Healthcare ser­vices Business and repair ser­vices (e.g., advertising, computer ser­vices) Transportation, communication, and utilities Financial ser ­vices Other professional and social ser­vices (e.g., engineering, accounting, computer) Business and repair ser­vices (e.g., advertising, computer ser­vices) Agriculture, mining, and manufacturing Education ser­vices Retail and ­wholesale trade Transportation, communication, and utilities Transportation, communication, and utilities Education ser­vices Financial ser ­vices Construction ser­vices Other professional and social ser­vices (e.g., engineering, accounting)

Share of increase in top 1% population, 1980–2015 40% 17% 15% 14% 7% 6% 6% 5% 4% 4% 4% 4% 3% 3% 2% 2% 1% 1% 1% 1%

Source: Author analy­sis of IPUMS USA. See replication code and related files for how categories ­were created using “ind1990” and “occ1990.” The column shows the change for each occupation-­ sector group in the number of workers in the top 1% from 2015 to 1980 divided by the total U.S. change in the number of workers in the top 1%.

310 • Appendix

International Analy­sis

Another approach to testing dif­f er­ent theories on what predicts the level and rise in income in­equality is to look across countries. This has the advantage of allowing for comparisons across dif­fer­ent institutional regimes and countries with dif­fer­ent characteristics and laws. I compiled 61 variables that I classified into nine dif­fer­ent theoretical viewpoints. L ­ ater, I w ­ ill discuss some of the individual variables in each. The evidence suggests that in­equality is closely linked to racial diversity, mea­sured as the share of population from vari­ous continents. The supplemental materials discuss the sources in more detail. Given that races do not differ by their inherent capacity for work, as I argued in detail in the preceding chapters, the fact that racial diversity predicts in­equality is likely the result of discrimination, segregation, and other ways in which racial minorities are cut off from full participation in markets and denied full access to high-­ quality public ser­vices. Demographic structure is also predictive of more in­equality, particularly if the working-­age population is large relative to the retirement-­age population, but we w ­ ill see this result is not robust. What I refer to as po­liti­cal economy variables perform moderately well on average, though some of the individual mea­sures are among the top performers. ­These mea­sure the structure of earnings through the lens of occupations and industries. A summary mea­sure of regulations on professional occupations that limit competition predicts greater in­equality. The capital-­versus-­labor variables perform rather poorly. Minimum-­wage regulations, ­union membership, the capital-­labor share of income, and regulations on firing workers are not well correlated with in­equality. The same is true for globalization mea­sures, which include the share of the population that is foreign born, how it has changed, and mea­sures of openness to trade. Institutional quality, skill, technological innovation—­ measured by patenting—­and the overall level of economic development all predict lower income in­equality and lower growth in­equality. This makes it unlikely that skill-­based technological change is a serious driving force of in­equality. Countries with more educated and inventive citizens, who have higher cognitive ability, tend to have low in­equality (­table A.6).

Appendix • 311

­ able A.6.  Correlations with level and change in in­equality by T theoretical concept

Racial diversity Demographic structure Po­liti­cal economy Capital vs. ­labor Globalization Institutional quality Skill Technology Level of development

Correlation with level of income in­equality

Correlation with change in income in­equality, 1980–2016 approx.

0.78 0.48 0.22 −0.15 −0.17 −0.30 −0.35 −0.36 −0.42

0.18 0.07 0.09 −0.13 −0.10 −0.04 0.06 −0.11 −0.20

Source: See online supplemental materials for details on sources and methods: https://github.com​ /jtrothwell.

The individual variable with the highest absolute value correlation with in­equality within the OECD is my preferred mea­sure of professional elite economic power: the 90th percentile of earnings for professional workers relative to the national median. This has an incredibly high correlation of 0.91 with the level of in­equality and 0.32 with the change in in­equality (­table A.7). Racial diversity is also highly predictive, as noted, followed by another mea­sure of elite-­professional dominance (median salary of professionals divided by median salary of all workers). Managerial dominance is also highly correlated, but the relationship is weaker than the professional-­worker relationship. Countries with larger tax revenue as a share of GDP—­which may indicate their commitment to high-­quality public services—­have lower in­ equality t­ oday and have seen lower growth in in­equality. ­These correlations should not be interpreted as causal, since omitted-­ variables bias is prevalent in all of them, but they do provide a baseline sense of what plausible causal relationships might look like. To further investigate, I regressed in­equality on the list of 61 variables, while controlling for racial diversity, the percentage of ­children raised by single parents, the patenting rate, the tertiary degree attainment rate, a trade freedom index, and the mean income, adjusted for purchasing power. The

­ able A.7.  Correlations with level and change in in­equality by T theoretical concept

Correlation with level of income in­equality 90th percentile of income of professional occupations/ Median income of all workers 18–65 Racial diversity of country Medium income of professional occupations/ Median income of all workers 18–65 90th percentile of income of managerial occupations/ Median income of all workers 18–65 Tax revenue as share of GDP Number of p­ eople of working age per number of retirement age Po­liti­cal stability and absence of violence/ terrorism Number of physicians with income in top 1%/total number of workers in top 1% Mean test scores, 15-­year-olds Medium income of managerial occupations/ Median income of all workers 18–65

Number of countries for level calculation

Correlation with change in income in­equality, 1980–2016 approx.

Number of countries for change calculation

0.91

20

0.32

20

0.78 0.75

35 20

0.18 0.22

35 20

0.73

21

−0.02

21

−0.72 0.71

35 33

−0.20 −0.12

35 33

−0.69

35

−0.14

35

0.67

8

0.37

8

−0.67

34

−0.03

34

0.64

21

−0.08

21

Note: See online supplemental materials for details on s​ ources a​ nd methods. These materials are available at https://​g ithub.com​/jtrothwell​/Republic_Equals.

Appendix • 313

sources are documented in the supplemental materials (https://­github​.­com​ /­jtrothwell​/­Republic​_ ­Equals). I used three mea­sures of in­equality: the UN Gini coefficient for the latest available year; the top-­one-­percent share of income from the World In­equality Database, and a composite mea­sure of the UN Gini, the WID top-­one-­ percent share, and the Luxembourg Income Study Gini coefficient. This index should capture multiple related dimensions of in­equality, given the dif­ fer­ent sources, and offers a robust summary mea­sure. In the OECD, only Chile, Mexico, and Turkey are more unequal than the United States, which is 1.7 standard deviations above the mean on the average of its three in­equality mea­sures. Israel is close ­behind at 1.2. Denmark has the lowest in­equality at −1.1 standard deviations below the mean. Notably, racial diversity and elite professional power are much more prevalent in the most unequal countries. For example, in Denmark, professionals at the 90th percentile are paid only two times as much (100 ­percent more) as the median worker, but in the United States they are paid 3.5 times as much (250 ­percent more). Only Mexico and Israel exhibit greater economic advantage for professionals than the United States (­table A.8). The regression results (which are available in the supplemental materials) reveal only four variables to be consistently significant (5 ­percent p-­values or lower) across the models: racial diversity (higher in­equality), patenting (lower in­equality), elite professional power (higher in­equality), and tax burden as a share of GDP (lower in­equality). The following variables are all consistently insignificant: the l­abor share of income, the change in the ­labor share of income, the number of ­union members as a share of all workers, changes in u­ nionization, the minimum wage, ­labor protections, the skill of the workforce, the gap in skills between top-­scoring students and the median student, the corporate tax rate, the individual tax rate, the progressivity of individual taxation, institutional quality as mea­sured by the Heritage Foundation and World Bank, trade openness, and exposure to globalization. ­These correlations and regression results do not provide direct evidence of a causal relationship. I rely on the logic of the book—­and the history and form of the regulations responsible—­rather than country-­level statistical analy­sis to make that case. I make all the data publicly available. In short, the results are strongly consistent with the main theme of this book: Providing high-­quality public services—­funded by government—­and

­ able A.8.  Summary data on in­equality, racial diversity, and elite T professional power

Chile Mexico Turkey United States Israel Estonia United Kingdom Greece Canada Poland Portugal Latvia Australia Japan ­Korea Spain Italy Ireland Germany France New Zealand Switzerland Hungary Austria Luxembourg Belgium Czech Republic Slovakia Finland Norway Netherlands Sweden Iceland Slovenia Denmark

In­equality index

UN Gini

3.2 3.0 2.0 1.7 1.2 0.5 0.5 0.5 0.4 0.3 0.2 0.2 0.2 0.1 0.1 0.1 0.0 0.0 −0.1 −0.2 −0.2 −0.3 −0.5 −0.6 −0.6 −0.6 −0.7 −0.8 −0.9 −0.9 −0.9 −1.0 −1.0 −1.0 −1.1

53.3 50.8 42.4 39.9 39.4 36.5 34.5 36.3 34.2 34.8 38.2 35.1 37.4 36.3 33.0 34.1 34.2 35.1 30.2 31.4 35.6 30.3 27.6 30.0 30.0 30.1 29.9 29.2 29.2 29.3 29.8 25.5 28.8 26.7 27.8

Top 1% share of income

21% 20% 13% 14% 12% 10% 9% 10% 12% 9% 9% 11% 13% 11% 8% 11% 10%

9% 7% 8% 6% 9% 6%

Racial diversity 0.51 0.52 0.20 0.43 0.46 0.02 0.10 0.00 0.32 0.01 0.03 0.05 0.20 0.00 0.00 0.02 0.00 0.00 0.06 0.13 0.31 0.03 0.07 0.03 0.02 0.04 0.00 0.01 0.00 0.00 0.08 0.01 0.00 0.00 0.02

Elite professional power 6.0 3.5 4.5 3.1 3.1 2.9 3.1

3.2 3.0 3.3 3.3 2.4 2.7 2.7 3.0

2.2 2.3 2.1 2.3 2.0

Note: Elite professional power is measured as the 90th percentile of earnings for professional workers relative to the national median. Racial diversity is the share of population from each continent. See online supplemental materials for details on sources and methods, available at https://github.com​ /jtrothwell/Republic_Equals.

Appendix • 315

curbing the power of elite professionals ­will lower income in­equality. Racial diversity predicts higher in­equality only b­ ecause prejudice and the doctrine of white supremacy have led to the exclusion of black minorities and Native Americans from public ser­vices and markets. The patent finding suggests that innovation lowers in­equality, but it could also be the case that egalitarian socie­ties are more innovative. I believe both are true. Combating racism and curbing professional privilege ­will sap the motivation and support for the unfair laws and be­hav­ior that generate and perpetuate in­equality. To conclude the arguments, markets that allow mutually beneficial exchange do not naturally lead to extreme in­equality. They tend to reward more productive p­ eople, but nearly every­one can contribute to society productively. The distribution of talent is naturally egalitarian across individuals and completely so across groups. This is not to deny that some p­ eople are born more conscientious and brighter than o­ thers, but the extra rewards generated by ­these advantages are modest relative to the enormous gaps in income observed in many democracies t­ oday.

NOTES

1. B ­ EHIND THE DISCONTENT 1. Time Staff, “­Here’s Donald Trump’s Presidential Announcement Speech,” Time, June 15, 2016, http://­time​.­com​/­3923128​/­donald​-­trump​-­announcement​-­speech​/­. 2. Cas Mudde, “The Far Right in the 2014 Eu­ro­pean Elections: Of Earthquakes, Cartels, and Designer Fascists,” Washington Post, May 30, 2014, https://­w ww​.­washingtonpost​ .­com​/­news​/­monkey​-­cage​/­wp​/­2014​/­05​/­30​/­the​-­far​-­right​-­in​-­the​-­2014​-­european​-­elections​-­of​ -­earthquakes​-­cartels​-­and​-­designer​-­fascists​/­​?­utm​_­term​=­​.­774be9678793; Brian Wheeler, Paul Seddon, and Richard Morris, “Brexit: All You Need to Know about the UK Leaving the EU,” BBC News, May 10, 2019, http://­w ww​.­bbc​.­com​/­news​/­u k​-­politics​-­32810887. 3. Bryony Jones and Hillary Clarke, “Le Pen ­Faces Macron in Final Round of Presidential Election,” CNN, April  24, 2017, http://­w ww​.­cnn​.­com​/­2017​/­0 4​/­23​/­europe​/­f rench​ -­presidential​-­election​-­results​/­. 4. My analy­sis of data from Gallup World Poll, https://­w ww​.­gallup​.­com​/­analytics​/­232838​ /­world​-­poll​.­aspx (accessed May 29, 2019). 5. Mudde, “The Far Right in the 2014 Eu­ro­pean Elections.” 6. “France Fuel Protests: Who Are the ‘Gilets Jaunes’ (Yellow Vests)?” BBC News, December 6, 2018. 7. Simon Schutz, “Germany’s Far-­R ight AfD Party Now Polls Second,” National Public Radio, September 30, 2018, https://­w ww​.­npr​.­org​/­2018​/­09​/­30​/­652284976​/­germanys​-­far​ -­right​-­afd​-­party​-­now​-­polls​-­second. 8. Annual growth rates are weighted by population. Data are from Penn World T ­ ables 9.0; see Robert C. Feenstra, Robert Inklaar, and Marcel P. Timmer, “The Next Generation of the Penn World T ­ able,” American Economic Review 105.10 (2015): 3150–82, www​.­ggdc​.­net​/­pwt. 9. Th ­ ese numbers are disputed by two other economists: Gerald Auten and David Splinter, “Top 1% Income Shares: Comparing Estimates Using Tax Data,” AEA Papers & Proceedings, forthcoming 2019, https://­w ww​.­aeaweb​.­org​/­conference​/­2019​/­preliminary​/­paper​ /­k 3bn5Qak. I analyzed appendix data from this research posted by Splinter and note two ­things: First, the Auten-­Splinter mea­sure of fiscal income (which is essentially all taxable income from ­labor or capital) showed an increase in top one-­percent income shares of 9.9 ­percent to 21.8 ­percent from 1979 to 2014, which essentially matches data from the Congressional Bud­get Office, as well as Thomas Piketty, Emmanuel Saez, and Gabriel Zucman, “Distributional National Accounts: Methods and Estimates for the United States,” Quarterly

318  •  Notes to Chapter 1 Journal of Economics 133.2 (2017): 553–609. Second, the dispute comes in considering pretax nonfiscal income, which includes concepts such as imputed rent to homeowners, unrealized pension contributions, realized capital gains, corporate retained earnings, and under-­ reported income, all of which are much harder to mea­sure and impute to individual tax filers relative to fiscal income. Auten-­Splinter estimate a top one-­percent share of just 13.1 ­percent ­a fter including nonfiscal income. A ­ fter making vari­ous assumptions, their calculations imply that the top one p­ ercent’s share of nonfiscal income went from 10.3 ­percent in 1979 to just 3.7 ­percent, despite a doubling of fiscal income. I agree with Piketty, Saez, and Zucman that this is highly implausible and contradicts reasonable assumptions about underreported income. But even if Auten and Splinter are closer to the truth, every­one agrees that t­ here was a large increase in income in­equality and that the top one p­ ercent control roughly 20 ­percent of taxable income. 10. Piketty, Saez, and Zucman, “Distributional National Accounts.” ­Table B3 in appendix 2 shows that from 1980 to 2014, average pretax income growth was as follows for all tax units aged 20 and over (with income split equally for ­those filing jointly): 61 ­percent for total population, 1 ­percent for the bottom 50 ­percent, 42 ­percent for the m ­ iddle 50th–90th income earners, 121 ­percent for the top 10 ­percent, and 204 ­percent for the top 1 ­percent. This income concept includes all ­labor and capital income, as well as Social Security, retirement income, disability, and unemployment insurance income. It does not include government taxes and transfers. The Census Bureau collects self-­reported income data from t­ hese sources (excluding capital gains) and finds that median income for full-­time year-­round workers increased by only 3 ­percent for men and 34 ­percent for ­women from 1980 to 2017; see “­Table P-36. Full-­Time, Year-­Round All Workers by Median Income and Sex: 1955 to 2017,” https://­ www​.­census​.­gov​/­data​/­tables​/­time​-­series​/­demo​/­income​-­poverty​/­historical​-­income​-­people​ .­html (accessed May 13, 2019). 11. See appendix 2, t­ able E3, in Piketty, Saez, and Zucman, “Distributional National Accounts.” 12. Edward N. Wolff, “Household Wealth Trends in the United States, 1962 to 2016: Has Middle Class Wealth Recovered?,” NBER Working Paper No. 24085, Cambridge, MA: National Bureau of Economic Research, 2017. National Accounts data show the national savings rate is also at historical lows, much lower than in the 1960s and 1970s; see U.S. Bureau of Economic Analy­sis, “­Table  2.1. Personal Income and Its Disposition,” https://​ www.​bea.​gov​/ (accessed June 20, 2019). 13. See Appendix T ­ able “Composition of Bottom 90%, Bottom 50%, and M ­ iddle 40% Pre-­Tax Income Shares: Taxable vs. Tax-­E xempt Income,” in Thomas Piketty, Emmanuel Saez, and Gabriel Zucman, “Distributional National Accounts: Methods and Estimates for the United States,” Quarterly Journal of Economics 133.2 (2017): 553–609. 14. Piketty, Saez, and Zucman, “Distributional National Accounts.” 15. It should be noted ­these data go beyond tax rec­ords to impute income from nontaxable benefits, such as ­those from employers and the government, as well as home ownership; see Piketty, Saez, and Zucman, “Distributional National Accounts.” They are the best and most comprehensive mea­sures of income available, b­ ecause they combine survey data with detailed tax data and detailed national accounting data. 16. The four Gallup World Poll questions are as follows: “(1) Which one of ­these phrases comes closest to your own feelings about your ­house­hold’s income ­these days: living comfortably on pre­sent income, getting by on pre­sent income, finding it difficult on pre­sent income,

Notes to Chapter 1  •  319 or finding it very difficult on pre­sent income? (2) Are you satisfied or dissatisfied with your standard of living, all the t­ hings you can buy and do? (3) Right now, do you feel your standard of living is getting better or getting worse? (4) Right now, do you think that economic conditions in the city or area where you live, as a w ­ hole, are getting better or getting worse?” I standardized responses to each question—­coded so negative responses have higher values—­ and took the mean. The resulting financial anxiety index is highly correlated with confidence in government and approval of the president or prime minister, depending on the country. To establish this, I regress po­liti­cal satisfaction (mea­sured by e­ ither confidence in government or approval of leaders) on financial anxiety, the log of income, the quintile income rank, educational attainment binary variables, age, age2 , and age3, gender, foreign-­born status, and employment status in OECD countries, controlling for country-­fi xed effects and year-­fi xed effects, during the years 2013–2017. The sample size is 129,551. A one standard deviation increase in financial anxiety increases probability of lacking confidence by 18  percentage points and increases probability of disapproval of government by 16 percentage points. 17. Benjamin M. Friedman, “The Moral Consequences of Economic Growth,” Society 43.2 (2006): 15–22; Francis Fukuyama, “Dealing with In­equality,” Journal of Democracy 22.3 (2011): 79–89. 18. Daron Acemoglu, “Import Competition and the ­Great US Employment Sag of the 2000s,” Journal of ­Labor Economics 34.S1 (2016): S141–98. 19. David Autor, David Dorn, Gordon H. Hanson, and Jae Song, “Trade Adjustment: Worker-­Level Evidence,” Quarterly Journal of Economics 129.4 (2014): 1799–1860. According to the authors’ data, the average wage in manufacturing was roughly $38,000 around 1990. They say the cumulative effect over 16 years was 46 ­percent of that baseline amount, which is roughly $1,100 per year. 20. Jonathan T. Rothwell, “Cutting the Losses: Reassessing the Costs of Import Competition to Workers and Communities,” Working Paper, March 22, 2017, https://­ssrn​.­com​ /­abstract​=­2920188, or http://­d x​.­doi​.­org​/­10​. ­2139​/­ssrn​. ­2920188. 21. Christian Broda and John Romalis, In­equality and Prices: Does China Benefit the Poor in Amer­i­ca? (Chicago: University of Chicago Press, 2008). 22. See U.S. Bureau of L ­ abor Statistics, “­Table 1110. Deciles of Income before Taxes: Annual Expenditure Means, Shares, Standard Errors, and Coefficients of Variation,” Consumer Expenditure Survey, 2017, https://­w ww​.­bls​.­gov​/­cex​/­tables​.­htm. 23. Michael A. Clemens and Jennifer Hunt, The L ­ abor Market Effects of Refugee Waves: Reconciling Conflicting Results, No. w23433 (Cambridge, MA: National Bureau of Economic Research, 2017). For an opposing argument, see George J. Borjas, We Wanted Workers: Unraveling the Immigration Narrative (New York: W. W. Norton, 2016). 24. G. I. P. Ottaviano and G. Peri, “Rethinking the Effect of Immigration on Wages,” Journal of the Eu­ro­pean Economic Association 10.1 (2012): 152–97; National Academies of Sciences, Engineering, and Medicine, The Economic and Fiscal Consequences of Immigration (Washington, DC: The National Academies Press, 2017), doi: https://­doi​.­org ​/­10​.­17226​ /­23550. 25. Borjas, We Wanted Workers. 26. See T ­ able 5-1 in National Academies of Sciences, Engineering, and Medicine, The Economic and Fiscal Consequences of Immigration. 27. George J. Borjas, “Yes, Immigration Hurts American Workers; The Candidates Tell Drastically Dif­fer­ent Stories about Immigration. ­They’re Both Skipping Half the Truth,”

320  •  Notes to Chapter 1 Politico, October 2016, https://­w ww​.­politico​.­com​/­magazine​/­story​/­2016​/­09​/­trump​-­clinton​ -­immigration​-­economy​-­unemployment​-­jobs​-­214216. 28. National Academies of Sciences, Engineering, and Medicine, The Economic and Fiscal Consequences of Immigration, “­Table 8-12. 75-­Year Net Pre­sent Value Flows for Consolidated Federal, State, and Local Governments for Two ­Future Bud­get Scenarios,” and ­Table 8-13, which shows the fiscal effects compared with natives. 29. Raj Chetty, Nathaniel Hendren, Maggie R. Jones, and Sonya R. Porter, Race and Economic Opportunity in the United States: An Intergenerational Perspective, No. w24441 (Cambridge, MA: National Bureau of Economic Research, 2018). 30. The methods appendix describes this analy­sis, and external data files available online reproduce it. 31. Bernie 2020, Issues, “Fight for Working Families,” https://­berniesanders​.­com​/­issues​ /­fight​-­for​-­working​-­families​/­(accessed May 29, 2019). 32. Loukas Karabarbounis and Brent Neiman, “The Global Decline of the L ­ abor Share,” Quarterly Journal of Economics 129.1 (2013): 61–103. 33. Jared Bern­stein, “Why ­Labor’s Share of Income Is Falling,” New York Times, Economix Blog, September 9, 2013. 34. International L ­ abor Organ­ization, “Global Wage Report 2012/13: Wages and Equitable Growth” (Geneva: International ­Labour Office, 2013). 35. U.S. Census Bureau, “P23-212, 65+ in the United States: 2010” (Washington, DC: U.S. Government Printing Office, 2014). 36. OECD, “Annual Survey of Large Pension Funds and Public Pension Reserve Funds: Report on Pension Funds’ Long-­Term Investments,” 2005, http://­w ww​.­oecd​.­org​/­daf​/­fi n​ /­private​-­pensions​/­2015​-­Large​-­Pension​-­Funds​-­Survey​.­pdf. 37. Louise Story, “G.M.’s Unit for Investing to Attract Pay Scrutiny,” New York Times, August 13, 2009, https://­w ww​.­nytimes​.­com​/­2009​/­08​/­14​/­business​/­14pay​.­html. 38. Matthew Rognlie, “Deciphering the Fall and Rise in the Net Capital Share: Accumulation or Scarcity?” Brookings Papers on Economic Activity 2015.1 (2016): 1–69, 2. 39. OECD, “HM1.3 Housing Tenures,” OECD Affordable Housing Database, n.d.,  https://­w ww​.­oecd​.­org ​/­els​/­family​/­H M1​-­3​-­Housing​-­tenures​.­pdf (accessed April  17, 2019). 40. Josh Bivens and Lawrence Mishel, “Understanding the Historic Divergence between Productivity and a Typical Worker’s Pay: Why It ­Matters and Why It’s Real” (Washington, DC: Economic Policy Institute, 2015). 41. For a definition of foreign versus domestic corporate profits, see Douglas R. Fox et al., “Chapter 13 Corporate Profits,” in NIPA Handbook: Concepts and Methods of the U.S. National Income and Product Accounts,” U.S. Bureau of Economic Analy­sis, https://­w ww​.­bea​ .­gov​/­system​/­fi les​/­2019​-­05​/­Chapter​-­13​.­pdf (accessed May 29, 2019). 42. My analy­sis of data from appendix ­tables A2 and A2b in Piketty, Saez, and Zucman, “Distributional National Accounts.” Specifically, I added up the bottom 90 ­percent and top 10 ­percent shares of national income that come from equities. This equals the value of equities divided by national income. I then divided the top 1 ­percent share of national income from equities by the total to get 64 ­percent in 2014. As it happens, 21 ­percent goes to the top 90 to 99 ­percent, and the rest, 15 ­percent, goes to the bottom 90 ­percent, according to their data. 43. My analy­sis of data from appendix t­ ables A2, A2b, and A2c in Piketty, Saez, and Zucman, “Distributional National Accounts.”

Notes to Chapter 1  •  321 44. My analy­sis of data from appendix t­ able A2b in Piketty, Saez, and Zucman, “Distributional National Accounts.” Specifically, I divided the top 1 ­percent share of total income by the top 1 ­percent share of equity income in 1980 and 2014. 45. U.S. Internal Revenue Ser­vices, “U.S. Individual Income Tax: Personal Exemptions and Lowest and Highest Bracket Tax Rates, and Tax Base for Regular Tax,” https://­w ww​ .­irs​.­gov​/­statistics​/­soi​-­tax​-­stats​-­historical​-­table​-­23 (accessed May 29, 2019). For data on effective payments, see Piketty, Saez, and Zucman, “Distributional National Accounts,” Appendix ­Table TG1. 46. My analy­sis of raw data from the World In­equality Database’s STATA download tool, “wid,” which pulls in raw data from the website wid.world. 47. Richard V. Reeves, Dream Hoarders: How the American Upper ­Middle Class Is Leaving Every­one Else in the Dust, Why That Is A Prob­lem, and What to Do About It (Washington, DC: Brookings Institution Press, 2018). 48. David Autor, “Skills, Education, and the Rise of Earnings In­equality among the ‘Other 99 ­Percent,’  ” Science 344.6186 (2014): 843–51. See also Philip Oreopoulos and Uros Petronijevic, “Making College Worth It: A Review of the Returns to Higher Education,” ­Future of ­Children 23.1 (2013): 41–65, http://­w ww​.­jstor​.­org​/­stable​/­23409488. 49. Patricia Cohen, “ ‘Superstar Firms’ May Have Shrunk Workers’ Share of Income,” New York Times, March 8, 2017; David Autor, David Dorn, Lawrence Katz, Christina Patterson, and John van Reenen, “The Fall of the ­Labor Share and the Rise of Superstar Firms,” National Bureau of Economic Research Working Paper No. 23396 (2017). 50. David Card, “The Causal Effect of Education on Earnings,” in Handbook of ­Labor Economics, 3: 1801–63 (Elsevier, 1999). 51. N. Gregory Mankiw, “Defending the One P ­ ercent,” Journal of Economic Perspectives 27.3 (2013): 21–34. 52. For evidence of large industry effects on earnings, controlling for ­human capital, see John Abowd, Francis Kramarz, Paul Lengermann, Kevin McKinney, and Sébastien Roux, “Persistent Inter-Industry Wage Differences: Rent Sharing and Opportunity Costs,” IZA Journal of Labor Economics 1.1 (2012): 1–25. 53. Jason M. Thomas, “Where Have All the Public Companies Gone?,” Wall Street Journal, November 16, 2017; Ian Hathaway and Robert E. Litan, “The Other Aging of Amer­i­ca: The Increasing Dominance of Older Firms” (Washington, DC: The Brookings Institution, 2014). 54. Christina Starmans, Mark Sheskin, and Paul Bloom, “Why P ­ eople Prefer Unequal Socie­ties,” Nature H ­ uman Behaviour 1 (2017): 0082. 55. Jon Bakija, Adam Cole, and Bradley T. Heim, “Jobs and Income Growth of Top Earners and the ­Causes of Changing Income In­equality: Evidence from US Tax Return Data,” unpublished manuscript, Williams College, 2012. 56. Luxembourg Income Study (LIS) Database, 2019, http://­w ww​.­lisdatacenter​.­org. 57. Sherwin Rosen, “The Economics of Superstars,” American Economic Review 71.5 (1981): 845–58; Autor et al., “The Fall of the ­Labor Share and the Rise of Superstar Firms.” 58. World Trade Organ­ization, “Trade and Tariffs,” n.d., https://­w ww​.­wto​.­org ​/­english​ /­thewto​_­e​/­20y ​_­e​/­wto​_­20​_­brochure​_­e​.­pdf (accessed April 17, 2019). 59. This is the population-­weighted average share of workers in the top 1 ­percent of income earners in the mining, manufacturing, transportation, storage, and communications sectors for 24 OECD countries with available data. Data are from the Luxembourg Income Study (2019), using its sector classifications. The countries included use data from 2010 or ­later with the exception of Sweden, where the latest year is 2005.

322  •  Notes to Chapter 1 60. Using data from the U.S. Bureau of Economic Analy­sis, this calculation divides exports of financial ser­v ices (roughly $100 billion) by the value added by the financial and insurance ser­vices sector for 2015 ($1.3 trillion). Manufacturing exports, valued at just over $1.1 trillion, and divided by GDP, $2.2 trillion, ­were calculated from the United States International Trade Commission DataWeb. 61. This relationship holds using the World Income Database (https://­wid​.­world), which is from tax rec­ords, or the Luxembourg Income Study, which is from survey-­based sources. The correlation coefficient with the healthcare, education, and public administration share of the one p­ ercent using “GiniAll” from the latter database is 0.63 for 19 countries. For changes, it is 0.68 for 13 countries. This is calculated using Gini coefficient changes from 1979–1981 to 2008–2014, averaging within each time period. See appendix for more information. 62. Japan has the lowest share of the one ­percent from ­these domestic ser­vice industries, but not b­ ecause it has a high percentage of top earners in global industries. Rather, Japan has a disproportionate share of top earners in other domestic industries such as construction (14 ­percent) and unclassifiable ser­vices (16 ­percent). 63. Analy­sis of 2015 American Community Survey via IPUMS USA, University of Minnesota, www​.­ipums​.­org. Steven Ruggles, Sarah Flood, Ronald Goeken, Josiah Grover, Erin Meyer, Jose Pacas, and Matthew Sobek, IPUMS USA: Version 9.0 [dataset] (Minneapolis, MN: IPUMS, 2019; https://­doi​.­org​/­10​.­18128​/­D010​.­V9​.­0). 64. Lawrence Mishel and Natalie Sabadish, “CEO Pay and the Top 1%: How Executive Compensation and Financial-­Sector Pay Have Fueled Income In­equality,” EPI Issue Brief 331 (2012). 65. Analy­sis of 2015 American Community Survey via IPUMS USA. 66. Th ­ ese figures use the latest available data from the Luxembourg Income Study (2019), using the most-detailed occupational categories. 67. It’s also worth pointing out that most man­a g­ers in the top 1 ­percent work in domestic sectors. In the United States, only 28 ­p ercent of man­a g­ers in the top 1 ­p ercent work in the export or tech-­oriented sectors of agriculture, mining, manufacturing, and communications. 68. Analy­sis of 2015 American Community Survey via IPUMS USA. 69. 2011 data from Statistics Canada via Minnesota Population Center, Integrated Public Use Microdata Series, International: Version 7.1 [dataset] (Minneapolis, MN: IPUMS, 2018. https://­doi​.­org​/­10​.­18128​/­D020​.­V 7​.­1). 70. Electronic Health Rec­ords Incentives Program, Centers for Medicare and Medicaid Ser­ vices, https://­w ww​.­cms​.­gov​/­Regulations​-­and​-­Guidance​/­Legislation​/­EHRIncentive­Programs​ /­index​.­html​?­redirect​=­​/­ehrincentiveprograms/ (accessed June  20, 2017). As late as 2007, two-­thirds of physician’s offices did not use electronic medical rec­ords; the share of physicians in the top 1 ­percent has not substantially changed since use became more widespread; see Chun-­Ju Hsaio and Esther Hing, “Trends in Electronic Health Rec­ord System Use among Office-­Based Physicians: United States, 2007–2012,” National Health Statistics Report 75 (May 20, 2014), https://­w ww​.­cdc​.­gov​/­nchs​/­data​/­nhsr​/­nhsr075​.­pdf#x2013. 71. The OECD’s Programme for the International Assessment of Adult Competencies (PIAAC) survey collected cognitive per­for­mance data from approximately 5,000 U.S. residents in 2012. I excluded occupational categories with fewer than 50 respondents. Among the remainder, the highest-scoring occupational group was information and communications

Notes to Chapter 1  •  323 technology (ICT) professionals, who scored .95 standard deviations above the mean U.S. score for combined results across three subjects (literacy, numeracy, and problem-­solving). Science and engineering professionals’ scores ­were the next highest (0.88 standard deviations above the mean). Business and administrative professions, l­ egal, social, and cultural professions, and health professions all scored lower: 0.76, 0.68, and 0.54, respectively. Teaching professionals scored 0.58. Respondents ­were also asked to report their “level of computer use”: 66  ­percent of ICT professionals reported complex computer use (as opposed to straightforward or moderate), compared with 27 ­percent of science and engineering professionals, 9 ­percent of business professionals, 3 ­percent of ­legal, social, and cultural professionals, and just 1 ­percent of health professionals. 72. Jonathan Rothwell, “Still Searching: Job Vacancies and STEM Skills” (Washington, DC: The Brookings Institution, 2014). 73. The U.S. total reported ­here adds ­legal occupations to arts, design, entertainment, sports, and media and social ser­vice and community occupations. 74. Luxembourg Income Study (2109), most recent year of data by country. 75. Autor et al., “The Fall of the ­Labor Share and the Rise of Superstar Firms.” 76. Another, more sophisticated approach takes into account the amount the com­pany spends on t­ hings like machines and energy, which complement workers. This is called “multifactor productivity” or “total-­factor productivity.” Multi­factor productivity and worker productivity are highly correlated, at least in the United States, and I use worker productivity ­because of the data challenges of mea­sur­ing multifactor productivity across countries and sectors. 77. Tesla, n.d. https://­w ww​.­tesla ​.­com/ (accessed June 19, 2017). 78. Technically, the key ratio is quality to cost, not price, but I wrote price above to simplify the discussion. In textbook microeconomic theory, prices and costs converge, but in real life this is not always the case, especially in the short term or if an industry is not very competitive. For example, a com­pany could keep prices the same but reduce costs by shifting its supply chain to a lower-cost provider of the same quality. This would still result in economic growth, but the gains would go to the com­pany’s o­ wners, not its customers. 79. The list is: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden, United Kingdom. Data analyzed ­here are from Eurostat National Accounts data, https://­ec​.­europa​.­eu​/­eurostat​/­data​/­database; productivity is calculated by dividing inflation-­adjusted gross value added by the number of employees for each industry. Data for the United States is from the Bureau of Economic Analy­sis, https://www.bea.gov/, using the same formula. 80. One explanation for rising returns to capital is the large and growing foreign trade deficits run by the United States. Trade deficits mean that foreign banks accumulate U.S. dollars and are essentially forced to find something to do with them, often buying U.S. securities or, during the housing b­ ubble of the early 2000s, purchasing mortgage backed securities. 81. Analy­sis of real GDP and employment data from Statistics Canada, https://­w ww150​ .­statcan​.­gc​.­ca​/­t1​/­tbl1​/­en​/­tv​.­action​?­pid​=­3610020701. Total GDP per worker growth in Canada over the same 2000–2015 period was 10 ­percent. Healthcare experienced −12 ­percent growth, whereas education experienced 4 ­percent growth. Professional ser­v ices saw just 1 ­percent growth. Manufacturing registered 16 ­percent growth. 82. Jonathan Rothwell, “The Declining Productivity of Education,” Washington, DC: The Brookings Institution, 2016.

324  •  Notes to Chapter 1 83. Jonathan Rothwell, “No Recovery: An Analy­sis of Long-­Term U.S. Productivity Decline” (Washington, DC: Gallup, 2016). 84. Kristopher J. Hult, Sonia Jaffe, and Tomas J. Philipson, How Does Technological Change Affect Quality-­Adjusted Prices in Health Care? Systematic Evidence from Thousands of Innovations, No. w22986 (Cambridge, MA: National Bureau of Economic Research, 2016). 85. Robert J. Gordon, The Rise and Fall of American Growth: The US Standard of Living Since the Civil War (Prince­ton, NJ: Prince­ton University Press, 2017). 86. My analy­sis of data from “Confidence in Institutions,” Gallup Poll, https://­news​ .­gallup​.­com​/­poll​/­1597​/­Confidence​-­Institutions​.­aspx (accessed May 31, 2019). 87. Sarah Flood, Miriam King, Renae Rod­gers, Steven Ruggles, and J. Robert Warren, “Integrated Public Use Microdata Series, Current Population Survey: Version 6.0 [dataset]” (Minneapolis, MN: IPUMS, 2018), https://­doi​.­org​/­10​.­18128​/­D030​.­V6​.­0. 88. Benjamin I. Page, Larry M. Bartels, and Jason Seawright, “Democracy and the Policy Preferences of Wealthy Americans,” Perspectives on Politics 11.1 (2013): 51–73. 89. Centers for Medicare and Medicaid Ser­vices, The National Health Expenditure Accounts, 2018, https://­w ww​.­cms​.­gov​/­Research​-­Statistics​-­Data​-­and​-­Systems​/­Statistics​-­Trends​ -­and​-­Reports​/ ­NationalHealthExpendData​/ ­NationalHealthAccountsHistorical​.­html. 90. My analy­sis of Gallup Po­liti­cal/Economic Daily track data, asked January 9–10, 2017. 91. Jonathan Rothwell, “The Minuscule Importance of Manufacturing in Far-­R ight Politics,” New York Times, Upshot, September 15, 2017. 92. Jonathan T. Rothwell and Pablo Diego-­Rosell, “Explaining Nationalist Po­liti­cal Views: The Case of Donald Trump,” November 2, 2016, https://­ssrn​.­com​/­abstract​=­2822059, or http://­d x​.­doi​.­org​/­10​.­2139​/­ssrn​.­2822059. 93. Eu­ro­pean Social Survey (ESS) Round 8 Data (2016), data file edition 2.1. NSD (­Norwegian Centre for Research Data), Norway, D ­ ata Archive and distributor of ESS data for ESS European Research Infrastructure Consortium (ERIC), https://www.euro​ peansocialsurvey.org/data/conditions_of_use.html. 94. Frank Newport, “Republicans Upgrade Views of Their Own Financial Situation,” Gallup Poll, March 22, 2017, https://­news​.­gallup​.­com​/­poll​/­206792​/­republicans​-­upgrade​ -­v iews​-­own​-­fi nancial​-­situation​.­a spx; Frank Newport, “Americans View Finances through Lens of Po­liti­cal Identity,” April 30, 2018, Gallup Poll, https://­news​.­gallup​.­com​/­poll​/­233576​ /­americans​-­view​-­finances​-­lens​-­political​-­identity​.­aspx (emphasis in original). 95. For this analy­sis, I updated my previous work by using zip code population data from the 2016 5-­Year American Community Survey. The sample size for this estimate is 73,950 white non-­Hispanic Americans randomly selected by phone in 2016. Rothwell and Diego-­ Rosell, “Explaining Nationalist Po­liti­cal Views.” 96. Results are based on author’s analy­sis of Eu­ro­pean Social Survey Round 7 Data (2014), data file edition 2.2. https://­w ww​.­europeansocialsurvey​.­org​/­. The question about immigration policy requests: “Please tell me how impor­tant you think each of ­these t­ hings should be in deciding ­whether someone born, brought up and living outside [country] should be able to come and live h ­ ere.” On a scale of 0 to 10, where 10 is “extremely impor­tant,” I coded p­ eople who answered that white race is impor­tant above 5. Questions about racial intelligence and work ethic w ­ ere: “Do you think some races or ethnic groups are born less intelligent than ­others?” and “Do you think some races or ethnic groups are born harder working than o­ thers?” In both cases, far-­right supporters w ­ ere more likely to say yes than ­those who did not vote for the far right by a statistically significant margin at 95  ­percent confidence intervals. Among far-­right supporters, 24  ­percent and

Notes to Chapter 1  •  325 47 ­percent answered yes to the question about intelligence and work ethic, respectively. Among ­those who did not vote for a far-­right party, only 14  ­percent and 40  ­percent, respectively, agreed with racist views. Far-­right party support was coded as 1 if the respondent indicated that he or she voted for a party that was described as far right in the ESS7 documentation. I further checked this for consistency against a useful summary analy­sis by Mudde, “The Far Right in the 2014 Eu­ro­pean Elections.” 97. Aaron Blake, “Republicans’ Views of Blacks’ Intelligence, Work Ethic Lag ­behind Demo­crats,’ ” Washington Post, March 31, 2017, https://­w ww​.­washingtonpost​.­com​/­news​/­the​ -­f ix​ /­w p​ /­2 017​/­0 3​ /­31​ /­t he​ -­g ap​ -­b etween​ -­republicans​ -­a nd​ -­d emocrats​ -­v iews​ -­o f​ -­a frican​ -­americans​-­just​-­hit​-­a​-­new​-­high​/­​?­utm​_­term​=­​.­e4d391c72273. 98. Blake, “Republicans’ Views of Blacks’ Intelligence.” 99. My analy­sis of Gallup data, conducted June 7–­July 1, 2016, from the Minority Rights and Relations poll. 100. John Sides, Michael Tesler, and Lynn Vavreck, Identity Crisis: The 2016 Presidential Campaign and the B ­ attle for the Meaning of Amer­i­ca (Prince­ton, NJ: Prince­ton University Press, 2018). 101. Louis Putterman and David N. Weil, “Post-1500 Population Flows and the Long-­ Run Determinants of Economic Growth and In­equality,” Quarterly Journal of Economics 125.4 (2010): 1627–82. 102. My analy­sis of Gallup data, conducted June 7–­July 1, 2016, for the Minority Rights and Relations poll. The sample included 2,314 non-­Hispanic black respondents who w ­ ere asked to answer yes or no to “Do you think racism against blacks is or is not widespread in the U.S.?” 103. ESS Round 8: Eu­ro­pean Social Survey Round 8 Data (2016). 104. Pew Research Center, “A Fragile Rebound for EU Image on Eve of Eu­ro­pean Parliamentary Elections,” May 2014, Washington, DC, https://www.pewresearch.org/wp-content​ /uploads/sites/2/2014/05/2014-05-12_Pew-Global-Attitudes-European-Union.pdf. 105. Pew Research Center, p. 27. 106. Alberto Alesina, Michela Carlana, Eliana La Ferrara, and Paolo Pinotti, Revealing Ste­reo­types: Evidence from Immigrants in Schools, No. w25333 (Cambridge, MA: National Bureau of Economic Research, 2018). 107. Alberto Alesina, Armando Miano, and Stefanie Stantcheva, Immigration and Re­ distribution, No. w24733 (Cambridge, MA: National Bureau of Economic Research, 2018). 108. Allum Bokhari and Milo Yiannopoulos, “An Establishment Conservative’s Guide to the Alt-­R ight,” Breitbart, March 29, 2016. 109. P. S. Forscher and N. Kteily, “A Psychological Profile of the Alt-­R ight,” Working Paper, August 10, 2017, https://­osf​.­io​/­preprints​/­psyarxiv​/­c9uvw. 110. Ashley Jardina, Sean McElwee, and Spencer Piston, “How Do Trump Supporters See Black P ­ eople? ‘Less Evolved,’ ” Slate, November 7, 2016, http://­w ww​.­slate​.­com​/­articles​ /­news​_­a nd​_ ­p olitics​/­p olitics​/­2 016​/­11​/­t he​_­majority ​_­of​_­t rump ​_­supporters​_­surveyed​ _­described​_­black ​_ ­people​_­as​_­less​.­html. 111. Leonie Huddy and Stanley Feldman, “On Assessing the Po­liti­cal Effects of Racial Prejudice,” Annual Review of Po­liti­cal Science 12 (February 6, 2009): 423–47. 112. Forscher and Kteily, “A Psychological Profile of the Alt-­R ight.” 113. My analy­sis of Gallup data, conducted June 7–­July 1, 2016, for the Minority Rights and Relations poll. Among non-­Hispanic whites, 40 ­percent of Trump supporters say the government should have no role in improving the social and economic position of black ­people. For non-­Hispanic white Republicans with an unfavorable view of Trump, the share

326  •  Notes to Chapter 2 is 29 ­percent, and it is 13 ­percent for non-­Hispanic white Demo­crats. The sample sizes are 531, 199, and 2,178, respectively. ­Those who lean ­toward one party ­were allocated to that party but the results are similar without leaners. 2. THE NATU­R AL FOUNDATIONS OF A JUST SOCIET Y 1. James Baldwin, Notes of a Native Son (Boston: Beacon Press, 1984), 13. 2. Anton-­Hermann Chroust and David L. Osborn, “Aristotle’s Conception of Justice,” Notre Dame Law Review 17.129 (1942): 129–43, http://­scholarship​.­law​.­nd​.­edu​/­ndlr​/­vol17​ /­iss2​/­2 . 3. The Hebrew god is quoted as commanding followers to use “just balances” and just currency. An unjust balance (or scale for mea­sur­ing weight) would presumably ­favor the merchant at the expense of the producer or customer, by not mea­sur­ing the true weight and undervaluing the market price of the product, as well as the materials and ­labor that went into it. Hence, to intentionally distort market prices is unjust. Even ­today, a common symbol of justice is a picture with a balanced scale. See Ezekiel 45:10, Bible, King James Version, https://­ quod​.­lib​.­umich​.­edu​/­cgi​/­k​/­kjv​/­kjv​-­idx​?­type​=­DIV2&byte​=­3276983; Leviticus 19:36, Bible, King James Version, http://­quod​.­lib​.­umich​.­edu​/­cgi​/­k​/­kjv​/­kjv​-­idx​?­type​=­DIV2&byte​=­497716. On conformity with law, Ezekiel 18 defines a just man as one who does what is “lawful and right,” and mentions following God’s laws as examples; Bible, King James Version, https://­ quod​.­lib​.­umich​.­edu​/­cgi​/­k​/­kjv​/­kjv​-­idx​?­t ype​=­DIV2&byte​=­3140652. 4. Donald E. Brown, “­Human Universals, ­Human Nature, and H ­ uman Culture,” Daedalus 133.4 (2004): 47–54. See Brown’s full list in Steven Pinker, The Blank Slate: The Modern Denial of ­Human Nature (New York: Penguin, 2003). 5. “Golden Rule,” Wikipedia​.­com, n.d., https://­en​.­wikipedia​.­org​/­wiki​/­Golden​_­Rule (accessed April 17, 2019). 6. Quoted in Abdusalam A. Guseinov, “The Golden Rule of Morality,” Rus­sian Social Science Review 55.6 (2014): 84–100; Ikechukwu Anthony Kanu, “Igwebuike as an Igbo-­A frican Ethic of Reciprocity,” Igwebuike: An African Journal of Arts and Humanities 3.2 (2017): 24. 7. Marcus Tullius Cicero, Ethical Writings of Cicero: De Officiis, De Senecture, De Amicitia, and Scorpio’s Dream (Veritatis Splendor Publications, 2014), 305, 507, 466 (Kindle ed.). 8. Marcus Tullius Cicero, “On Duties III,” in Selected Works, trans. Michael Grant (London: Penguin, 1960), 167. 9. Cicero, Ethical Writings. 10. While short of calling for the abolition of slavery, this is the most anti-­slavery sentiment I’ve been able to find anywhere in the ancient world, including in the words of the Jewish prophets, Jesus, and Muhammed, though all held generous attitudes t­ oward t­ hose with low social status. It is also worth pointing out that unlike in modern Amer­i­ca, Roman slavery was neither a hereditary condition in the ancient world nor even permanent for many slaves. A few former slaves even went on to be emperors. 11. Pope John Paul II, On Social Concern (Vatican City: Vatican Press, 1987), sec. 42, http://­w ww​.­vatican​.­va​/­holy ​_­father​/­john​_ ­paul ​_­i i​/­encyclicals​/­documents​/­h f​_ ­jp​-­i i ​_­enc​ _­30121987​_­sollicitudo​-­rei​-­socialis​_­en​.­html. 12. “The Teaching of Amenemhat I: Section 3,” Digital Egypt for Universities, n.d., http://­ www​.­ucl​.­ac​.­uk​/­museums​-­static​/­digitalegypt​/­literature​/­teachingaisec3​.­html (accessed May 2,

Notes to Chapter 2  •  327 2018). I am indebted to Dominic Perry’s “The History of Egypt Podcast” for this insight; https://­egyptianhistorypodcast​.­com​/­. 13. Sarah F. Brosnan and Frans B. M. de Waal, “Monkeys Reject Unequal Pay,” Nature 425.6955 (2003): 297–99. 14. Frans de Waal, “Moral Be­hav­ior in Animals,” TEDxPeachtree, November 2011, https://­w ww​ .­t ed​ .­c om​ /­t alks​ /­f rans​ _­d e ​ _­w aal ​ _­d o​ _ ­a nimals​ _ ­h ave ​ _­m orals​ /­t ranscript​ ?­language​=­en. 15. Frans B. M. de Waal, Kristin Leimgruber, and Amanda R. Greenberg, “Giving Is Self-­ Rewarding for Monkeys,” Proceedings of the National Acad­emy of Sciences 105.36 (2008): 13685–89. 16. J. M. Burkart, E. Fehr, C. Efferson, and C. P. van Schaik, “Other-­Regarding Preferences in a Non-­Human Primate: Common Marmosets Provision Food Altruistically,” Proceedings of the National Acad­emy of Sciences 104.50 (2007): 19762–66. 17. Juliane Bräuer and Daniel Hanus, “Fairness in Non-­Human Primates?” Social Justice Research 25.3 (2012): 256–76. 18. Ernst Fehr and Simon Gächter, “Cooperation and Punishment in Public Goods Experiments,” American Economic Review 90.4 (2000): 980–94. 19. Samuel Bowles and Herbert Gintis, A Cooperative Species: H ­ uman Reciprocity and Its Evolution (Prince­ton, NJ: Prince­ton University Press, 2011). 20. Ernst Fehr and Simon Gächter, “Altruistic Punishment in ­Humans,” Nature 415.6868 (2002): 137–40. 21. Joseph Henrich, Robert Boyd, Samuel Bowles, Colin Camerer, Ernst Fehr, Herbert Gintis, and Richard McElreath, “In Search of Homo Economicus: Behavioral Experiments in 15 Small-­Scale Socie­ties,” American Economic Review 91.2 (2001): 73–78; P. R. Blake, K. McAuliffe, J. Corbit, T. C. Callaghan, O. Barry, A. Bowie, L. Kleutsch, et al., “The ­Ontogeny of Fairness in Seven Socie­ties,” Nature 528.7581 (2015): 258–61. 22. Bowles and Gintis, A Cooperative Species. 23. Grit Hein, Yosuke Morishima, Susanne Leiberg, Sunhae Sul, and Ernst Fehr, “The Brain’s Functional Network Architecture Reveals ­Human Motives,” Science 351.6277 (2016): 1074–78. 24. Amy Dawel, Richard O’Kearney, Elinor McKone, and Romina Palermo, “Not Just Fear and Sadness: Meta-­Analytic Evidence of Pervasive Emotion Recognition Deficits for Facial and Vocal Expressions in Psychopathy,” Neuroscience & Biobehavioral Reviews 36.10 (2012): 2288–2304. 25. Abigail A. Marsh, Sarah A. Stoycos, Kristin M. Brethel-­Haurwitz, Paul Robinson, John W. VanMeter, and Elise M. Cardinale, “Neural and Cognitive Characteristics of Extraordinary Altruists,” Proceedings of the National Acad­e my of Sciences 111.42 (2014): 15036–41. 26. K. E. Langergraber, J. C. Mitani, and L. Vigilant, “The L ­ imited Impact of Kinship on Cooperation in Wild Chimpanzees,” Proceedings of the National Acad­e my of Sciences 104.19 (2007): 7786–90. 27. Bowles and Gintis, A Cooperative Species. 28. Katherine McAuliffe, Natalie Shelton, and Lauren Stone, “Does Effort Influence Inequity Aversion in Cotton-­Top Tamarins (Saguinus oedipus)?” Animal Cognition 17.6 (2014): 1289–1301; Sarah F. Brosnan and Frans B. M. de Waal, “Evolution of Responses to (Un)Fairness,” Science 346.6207 (2014): 1251776.

328  •  Notes to Chapter 2 29. Edward O. Wilson, The Social Conquest of Earth (New York: W. W. Norton, 2012), 4142 (Kindle ed.). 30. Jung-­Kyoo Choi and Samuel Bowles, “The Coevolution of Parochial Altruism and War,” Science 318.5850 (2007): 636–40. 31. My analy­sis of 61 hunter-­gatherer cultures from the “Standard Cross-­Cultural Sample” shows that only four w ­ ere reported as not conducting intergroup trade; data are from George P. Murdock and Douglas R. White, “Standard Cross-­Cultural Sample,” Ethnology 8.4 (1969): 329–69. 32. Fotini Christia, Alliance Formation in Civil Wars (Cambridge: Cambridge University Press, 2012). 33. Thomas F. Pettigrew and Linda R. Tropp, “A Meta-­A nalytic Test of Intergroup Contact Theory,” Journal of Personality and Social Psy­chol­ogy 90.5 (2006): 751; Jonathan T. Rothwell, “The Effects of Racial Segregation on Trust and Volunteering in US Cities,” Urban Studies 49.10 (2012): 2109–36; Marcus Alexander and Fotini Christia, “Context Modularity of ­Human Altruism.” Science 334.6061 (2011): 1392–94. 34. Cicero, Ethical Writings, 389, 383. 35. Plato, The Republic, trans. Benjamin Jowett (Titan Classics, 2015), 989 (Kindle ed.). 36. Plato, The Republic, 349. 37. Plato, Laws, in Plato: The Complete Works (e-­artnow, 2015), 42816 (Kindle ed.). 38. Plato, The Republic, 989. 39. Steven Michels, “Democracy in Plato’s Laws,” Government, Politics, and Global Studies Faculty Publications, Paper 24, 2004, http://­d igitalcommons​.­sacredheart​.­edu​ /­gov​_­fac​/­24. 40. First, Sparta initiated a rule by two rather than just one king. Then, Sparta make the kings equal in power to a parliamentary council of 28 elders “in the most impor­tant m ­ atters.” Third, Sparta enacted some mechanism of electing “Ephors” who seemed to hold considerable legislative power. More impor­tant is that in laying out his ideal city, Plato departs somewhat from The Republic in advocating for a mix of monarchy and democracy, suggesting that the best attributes of Persia’s monarchy—­under a wise king such as Cyrus the ­Great—­should be combined with Athenian democracy. 41. Michels, “Democracy in Plato’s Laws”; Plato, Laws. 42. Plato, The Republic. 43. John Rawls, Justice as Fairness: A Restatement (Cambridge, MA: Harvard University Press, 2001), 42–43. 44. Kristin F. Butcher, Assessing the Long-­Run Benefits of Transfers to Low-­Income Families (Washington, DC: Brookings Institution Hutchins Center Working Paper 26, 2017), https://­w ww​.­brookings​.­edu​/­research​/­a ssessing​-­the​-­long​-­r un​-­benefits​-­of​-­transfers​-­to​-­low​ -­income​-­families​/­. 45. Cicero, Ethical Writings, 350. 46. Leif Wenar, “John Rawls,” in The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, 2017, https://­plato​.­stanford​.­edu​/­archives​/­spr2017​/­entries​/­rawls. 47. Paul Ricoeur, The Just, trans. David Pellauer (Chicago: University of Chicago Press, 2003), 54. 48. John Rawls, A Theory of Justice (Cambridge, MA: Harvard University Press, 1971), 3. 49. Karl Marx, “Critique of the Gotha Program,” in Marx/Engels Selected Works (Moscow: Pro­g ress Publishers, 1970), 3:13–30.

Notes to Chapter 3  •  329 50. Lars J. Lefgren, David P. Sims, and Olga B. Stoddard, “Effort, Luck, and Voting for Re­distribution,” Journal of Public Economics 143 (November 2016): 89–97. 51. Philip Taubman, “Soviet Law Widens Private Business,” New York Times, November  20, 1986, https://­w ww​.­nytimes​.­com​/­1986​/­11​/­2 0​/­world​/­soviet​-­l aw​-­w idens​-­private​ -­business​.­html. 52. Timothy Garton Ash, The Magic Lantern: The Revolution of ’89 Witnessed in Warsaw, Budapest, Berlin and Prague (New York: Atlantic Books, 2014). 53. Rawls, Justice as Fairness. 54. Robert Nozick, Anarchy, State, and Utopia (New York: Basic Books, 2013); Friedrich August Hayek, The Fatal Conceit: The Errors of Socialism, vol. 1 (Chicago: University of Chicago Press, 2011). 55. R.  P. Saller, Personal Patronage ­under the Early Empire (Cambridge: Cambridge University Press, 2002). 56. William  V. Harris,  Ancient Literacy (Cambridge, MA: Harvard University Press, 1991). 57. Amartya Sen, Development as Freedom (Oxford: Oxford University Press, 2001). 58. Martha Nussbaum, “Capabilities as Fundamental Entitlements: Sen and Social Justice,” Feminist Economics 9.2–3 (2003): 33–59. 3. MERIT-­B ASED EGALITARIANISM 1. Bart de Langhe, Philip M. Fernbach, and Donald R. Lichtenstein, “Navigating by the Stars: Investigating the ­Actual and Perceived Validity of Online User Ratings,” Journal of Consumer Research 42.6 (2016): 817–83; Abhijit Gosavi, William Daughton, Ozge Senoz, and V. A. Samaranayake, “Consumer Perception of US and Japa­nese Automobiles: A Statistical Comparison via Consumer Reports and J. D. Power & Associates Data,” International Journal of Engineering Management and Economics 6.1 (2016): 1–18; Jose A. Guajardo, Morris A. Cohen, and Serguei Netessine, “Ser­vice Competition and Product Quality in the US Automobile Industry,” Management Science 62.7 (2016): 1860–77, doi: 10.1287/mnsc​.2015​. 2195. 2. Jonathan T. Rothwell, “Assessing the Validity of Consumer Ratings for Higher Education: Evidence from a New Survey,” October 20, 2017. https://­ssrn​.­com​/­abstract​=­2982395. 3. Patrick Kampkötter and Dirk Sliwka, “The Complementary Use of Experiments and Field Data to Evaluate Management Practices: The Case of Subjective Per­for­mance Evaluations,” Journal of Institutional and Theoretical Economics 172.2 (2016): 364–89. 4. Nicholas Bloom, Christos Genakos, Raffaella Sadun, and John Van Reenen, “Management Practices across Firms and Countries,” Acad­emy of Management Perspectives 26.1 (2012): 12–33. 5. Anders Frederiksen, Fabian Lange, and Ben Kriechel, “Subjective Per­for­mance Evaluations and Employee ­Careers,” Journal of Economic Be­hav­ior & Organ­ization 134 (2017): 408–29. 6. T. J. Kane and D. O. Staiger, “Gathering Feedback for Teaching: Combining High-­ Quality Observations with Student Surveys and Achievement Gains” (Seattle, WA: Bill and Melinda Gates Foundation, 2012); M. Caridad Araujo, Pedro Carneiro, Yyannú Cruz-­ Aguayo, and Norbert Schady, “Teacher Quality and Learning Outcomes in Kindergarten,” Quarterly Journal of Economics 131.3 (2016): 1415–53.

330  •  Notes to Chapter 3 7. Alan Manning, “We Can Work It Out: The Impact of Technological Change on the Demand for Low-­Skill Workers,” Scottish Journal of Po­liti­cal Economy 51.5 (2004): 581–608. 8. To mea­sure IQ, I used the percentile scores from a variety of dif­fer­ent IQ tests ­because subjects w ­ ere not all given the same test. The list is available on the NLSY website (https://­ www​.­nlsinfo​.­org​/­investigator​/­pages​/­search​.­jsp​?­s​=­NLSY79) and includes the Stanford-­Binet, Henmon-­Nelson, Wechsler, Lorge-­Thorndike, and a few ­others, all administered in 1979. 9. This is consistent with other evidence that crystalized intelligence is more impor­ tant than fluid intelligence in terms of predicting life outcomes, such as years of education. Mathilde Almlund, Angela Lee Duckworth, James J. Heckman, and Tim D. Kautz, “Personality Psy­chol­ogy and Economics,” in Handbook of the Economics of Education (Elsevier, 2014), 4: 1–181. 10. ­These results can be replicated using the NLSY 1997 database and including a squared term for the Armed Services Vocational Aptitude Battery (ASVAB) score that combines math and verbal per­for­mance; see https://­w ww​.­nlsinfo​.­org​/­investigator​/­pages​/­search​.­jsp​?­s​ =­NLSY79. 11. I supplement this with “Program for the International Assessment of Adult Competencies (PIAAC) 2012/2014: U.S. National Supplement Public Use Data Files—­Household,” a U.S. file not originally included in the PIAAC release. It is available through the U.S. National Center for Educational Statistics, https://­nces​.­ed​.­gov​/­pubsearch​/­pubsinfo​.­asp​?­pubid​ =­2016667REV. I remove the prison population to maintain a consistent population for comparison. 12. For a regression of log earnings (which includes salary, wages, and bonuses) on age, age2 , age3, and age4, mea­sured in five-­year bands, the adjusted r-­squared is 0.235. For a regression of log earnings on cognitive ability, it is 0.225. For educational attainment and for gender, the r-­squared is 0.23. Th ­ ese regressions include country-­fi xed effects. When combining all of the variables, the coefficient on cognitive ability is 0.13, and all the other terms remain highly significant. 13. To make this calculation, I regress the log of earnings on cognitive ability, educational attainment (mea­sured as binary variables for primary and secondary school completion and associate’s, bachelor’s, and gradu­ate degrees), age, age2 , age3, and age4, and an indicator for male for 72,000 individuals. I do the same for health status, mea­sured on a 1–5 scale. I then predict both outcomes based on the under­lying variables. The Gini coefficients for income and health are .47 and .52, respectively. The Gini coefficients for predicted income and predicted health (based on IQ and the other f­ actors) are .17 and .21, respectively. 14. My analy­sis of IPUMS CPS; Sarah Flood, Miriam King, Renae Rod­gers, Steven Ruggles, and J. Robert Warren, Integrated Public Use Microdata Series, Current Population Survey: Version 6.0 [dataset] (Minneapolis, MN: IPUMS, 2018), https://­doi​.­org​/­10​.­18128​ /­D030​.­V6​.­0. 15. To calculate this, I use $50,000 for the average salary in the PIAAC database (the mean is actually $41,000 in purchasing power parity adjusted dollars but would be higher in a­ ctual dollars ­because Eu­rope tends to be more expensive). I multiply the standardized IQ score (3.27 for IQ of 149) by .13, which is the average predicted effect of IQ on income, which yields 0.42. The natu­ral log of $50,000 is 10.8; 10.8 plus .42 yields 11.23, which amounts to $76,000. The 149 IQ figure for the top 0.1 ­percent comes from Jack Cox, “Smarter than 99.9% of the Rest of Us,” Denver Post, June 20, 2005, https://­w ww​.­denverpost​.­com​/­2005​/­06​/­20​/­smarter​ -­than​-­99​-­9​-­of​-­the​-­rest​-­of​-­us​/­. The average predicted effect comes from the regression described above.

Notes to Chapter 3  •  331 16. On the basis of the NLSY 1997, the Gini coefficients for personality, cognitive ability, income, and health are 0.22, 0.37, 0.41, and 0.49, respectively. 17. I calculate an age-­cohort-­adjusted mea­sure of IQ by combining PIAAC test scores for the math, literacy, and technology sections of the test; ages ­were grouped into 11 categories. I further removed the incarcerated population. The raw data with race variables included are available through the U.S. National Center for Educational Statistics, https://­nces​.­ed​.­gov​ /­pubsearch​/­pubsinfo​.­asp​?­pubid​=­2016667REV. 18. Richard J. Herrnstein and Charles Murray, The Bell Curve: Intelligence and Class Structure in American Life (New York: Simon and Schuster, 2010). 19. This NLS 1997 finding was generated by regressing the log of income in 2013 on the summary ASVAB score and binary variables for being male, having a m ­ other with a bachelor’s degree or higher education, black, Jewish, Asian, Hispanic, or multiracial, relative to a reference group of white non-­Hispanics. The coefficient on black is negative and significant, and the reverse is true for the coefficients on maternal education and male. The implication is that pay gaps would be reduced for females, black p­ eople, and t­ hose born with less-­educated ­mothers if pay w ­ ere based only on the ASVAB score. The finding from the PIAAC used a similar model but adjusted for region by including binary variables for the South, Midwest, and West (with the Northeast as reference) categories, and controlling for suburban, town, and rural, with city being the reference category. The variable for male and maternal education of bachelor’s degree or higher ­were statistically significant and positive. The variable for black was negative but not significant. 20. ­These data combine O*NET Version 19 scores on knowledge domains by occupation with American Community Survey data from the 2011–2015 sample, as produced and made available through IPUMS USA (University of Minnesota, www​.­ipums​.­org). I created the crosswalk between them using methods similar to ­those discussed in Jonathan Rothwell, “Defining Skilled Technical Work,” prepared for the National Academies Board on Science, Technology, and Economic Policy Proj­ect on “The Supply Chain for Middle-­Skilled Jobs: Education, Training and Certification Pathways,” September  1, 2015, https://­sites​.­national​ academies​.­org​/­cs​/­groups​/­pgasite​/­documents​/­webpage​/­pga​_­167744​.­pdf. 21. Lawrence H. Summers, “Remarks at NBER Conference on Diversifying the Science & Engineering Workforce,” Cambridge, MA, January 14, 2005, https://­w ww​.­harvard ​.­edu​ /­president​/­speeches​/­summers ​_ ­2005​/­nber​.­php. 22. National Science Foundation, National Center for Science and Engineering Statistics, “­Women, Minorities, and Persons with Disabilities in Science and Engineering,” special tabulations of U.S. Department of Education, National Center for Education Statistics, Integrated Postsecondary Education Data System, Completions Survey, ­table 7-2, “Doctoral Degrees Awarded to W ­ omen, by field: 2004–2014,” https://­w ww​.­nsf​.­gov​/­statistics​/­2017​ /­nsf17310​/­data​.­cfm. 23. Charlotte Witt and Lisa Shapiro, “Feminist History of Philosophy,” in The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta (2017), https://­plato​.­stanford​.­edu​/­archives​ /­spr2017​/­entries​/­feminism​-­femhist​/­. 24. Jonathan Wai, David Lubinski, and Camilla P. Benbow, “Spatial Ability for STEM Domains: Aligning over 50 Years of Cumulative Psychological Knowledge Solidifies Its Importance,” Journal of Educational Psy­chol­ogy 101.4 (2009): 817; Lloyd G. Humphreys, David Lubinski, and Grace Yao, “Utility of Predicting Group Membership and the Role of Spatial Visualization in Becoming an Engineer, Physical Scientist, or Artist,” Journal of Applied Psy­ chol­ogy 78.2 (1993): 250.

332  •  Notes to Chapter 3 25. PISA 2015 database, http://­w ww​.­oecd​.­org​/­pisa​/­data​/­2015database​/­. 26. Among 15-­year-­olds who score 2 standard deviations or higher (using international means and variation) in the United States, 41 ­percent are female for science, 37 ­percent for math, and 55 ­percent for reading. 27. To test this, I regressed science and math scores separately on a variable that is the gap between paternal and maternal education levels (dad-­mom) for girls and then for boys. A positive gap predicted lower scores for girls but higher scores for boys. The differences w ­ ere statistically significant in three of the four cases (science for girls and math and science for boys). 28. Michel Duyme, Annick-­Camille Dumaret, and Stanislaw Tomkiewicz, “How Can We Boost IQs of ‘Dull C ­ hildren’? A Late Adoption Study,” Proceedings of the National Acad­ emy of Sciences 96.15 (1999): 8790–94. 29. For comprehensive discussion of personality, cognitive ability, and some of the evidence on their determinants, see Almlund et al., “Personality Psy­chol­ogy and Economics.” 30. Almlund et al., “Personality Psy­chol­ogy and Economics.” 31. Frederiksen, Lange, and Kriechel, “Subjective Per­for­mance Evaluations and Employee ­Careers.” 32. Almlund et al., “Personality Psy­chol­ogy and Economics.” For discussion of the relationship between job per­for­mance and cognitive ability, see Frank L. Schmidt and John Hunter, “General ­Mental Ability in the World of Work: Occupational Attainment and Job Per­for­mance,” Journal of Personality and Social Psy­chol­ogy 86.1 (2004): 162–73. 33. Frank L. Schmidt, In-­Sue Oh, and Jonathan A. Shaffer, “The Validity and Utility of Se­lection Methods in Personnel Psy­chol­ogy: Practical and Theoretical Implications of 100 Years of Research Findings,” Fox School of Business Research Paper, October 17, 2016, https://­ssrn​.­com​/­abstract​=­2853669. 34. This lit­er­a­ture finds that conscientiousness has a negative correlation with IQ (General ­Mental Ability) of −0.07. The correlation between IQ and integrity is positive but only 0.05. Emotional stability is correlated with IQ at .16. The other “Big 5” personality traits (extroversion, openness to experience, and agreeableness) have very low correlations with per­for­mance, but some have incremental validity in that they add to the predictive power of se­lection when combined with IQ. 35. Schmidt, Oh, and Shaffer, “The Validity and Utility of Se­lection Methods in Personnel Psy­chol­ogy.” 36. Patrick F. McKay and Michael A. McDaniel, “A Reexamination of Black-­White Mean Differences in Work Per­for­mance: More Data, More Moderators,” Journal of Applied Psy­ chol­ogy 91.3 (2006): 538. 37. Philip Bobko, Philip L. Roth, and Denise Potosky, “Derivation and Implications of a Meta-­A nalytic Matrix Incorporating Cognitive Ability, Alternative Predictors, and Job Per­ for­mance,” Personnel Psy­chol­ogy 52.3 (1999): 561–89. 38. James K. Harter, Frank Schmidt, Sangeeta Agrawal, Stephanie Plowman, and Anthony Blue, “The Relationship between Engagement at Work and Orga­nizational Outcomes: 2016 Q12 Meta-­A nalysis: Ninth Edition” (Washington, DC: Gallup, 2016). 39. J. K. Harter, F. L. Schmidt, J. W. Asplund, E. A. Killham, and S. Agrawal, “Causal Impact of Employee Work Perceptions on the Bottom Line of Organ­izations,” Perspectives on Psychological Science 5.4 (2010): 378–89. 40. For information on how Gallup measures engagement, see Gallup Q12 Employee Engagement Survey, https://­q12 ​.­gallup​.­com​/­public​/­en​-­us​/­Features. Gallup survey sources used here are described at the following links: Gallup Panel, https://­w ww​.­gallup​.­com​

Notes to Chapter 4  •  333 /­analytics​/­213695​/­gallup​-­panel​.­aspx; Gallup Healthways, https://­w ww​.­gallup​.­com​/­175196​ /­gallup​-­healthways​-­index​-­methodology​.­aspx. 41. Th ­ ese gaps (.10 standard deviations) w ­ ere calculated by regressing a summary mean index of the 12 Q12 items ­a fter standardizing each one on dummy variables for each racial group and for gender, controlling for age, age-­squared, and age-­cubed. The sample was ­limited to workers, but relaxing that makes l­ ittle difference. The main finding was robust to controls for educational attainment and part-­time work status. Similar results ­were found using the larger Gallup Healthways Wellbeing Index sample of 85,000 workers, https://­w ww​.­gallup​ .­com​/­175196​/­gallup​-­healthways​-­index​-­methodology​.­aspx (accessed June 1, 2019). 42. This analy­sis regresses three items from the Q12 (standardized) on age (age squared and cubed), gender, racial binary variables, education variables, marital status, the presence of ­children in the home, and employment status (full-­time self-­employed and full-­time employed). I also control for ­whether the respondent reports that the com­pany is hiring or reducing staff. ­People working at firms in which employment is being reduced score 0.42 standard deviations lower on engagement; at growing firms, engagement is 0.19 standard deviations higher. The religious, racial, and gender differences are not substantially affected by including or removing ­these controls, but including a control for education does reduce the predicted Jewish score. 43. Similar results can be found using the three items from the larger survey. Since one Q12 item could be seen as directly mea­sur­ing per­for­mance, I examine that separately. On receiving praise for good work, 43 ­percent of black workers answered with a 4 or 5, suggesting agreement or strong agreement, compared to 44 ­percent of Asians, 52 ­percent of white ­people, and 55 ­percent of Hispanics. Again, this suggests small gaps that are inconsistent with typical social status perceptions of t­ hese groups. 44. Patrick F. McKay, Derek R. Avery, and Mark A. Morris, “Mean Racial-­Ethnic Differences in Employee Sales Per­for­mance: The Moderating Role of Diversity Climate,” Personnel Psy­chol­ogy 61.2 (2008): 349–74. 45. Kim Mai-­Cutler, “­Here’s a Detailed Breakdown of Racial and Gender Diversity Data across U.S. Venture Capital Firms,” TechCrunch, October 6, 2015, https://­techcrunch​.­com​ /­2015​/­10​/­06​/­s23p​-­racial​-­gender​-­diversity​-­venture​/­. 46. S. B. Badal and J. H. Streur, “Builder Profile 10: Methodology Report” (Washington, DC: Gallup, 2014), https://­w ww​.­gallupstrengthscenter​.­com​/­EP10​/­en​-­US​/­GetFile​ ?­fi leName​=­EP10%2FEP10​_­TalentDefinitions​.­pdf&language​=­en​-­US. 47. Jonathan Rothwell, “No Recovery: An Analy­sis of Long-­Term U.S. Productivity Decline” (Washington, DC: Gallup, 2016). 48. Pamela Villarreal, How Much Are Teachers ­Really Paid? A Nationwide Analy­sis of Teacher Pay (Madison, WI: McIver Institute, 2014). 4. THE IMPORTANCE OF EQUAL ACCESS TO PUBLIC GOODS AND MARKETS 1. Adult education is considered a quasi-­public good and should be partly subsidized with public money. It is partly private ­because t­ here are large private benefits to education that immediately prepares adults for a certain occupation or c­ areer path, in the form of higher pay, for example. But is also public, ­because every­one is more productive when they are allowed to specialize and exchange with other specialists.

334  •  Notes to Chapter 4 2. Earl J. Hamilton, “The Role of Mono­poly in the Overseas Expansion and Colonial Trade of Eu­rope before 1800,” American Economic Review 38.2 (1948): 33–53. 3. Sheilagh Ogilvie, “The Economics of Guilds,” Journal of Economic Perspectives 28.4 (2014): 169–92; Sheilagh Ogilvie, Institutions and Eu­ro­pean Trade: Merchant Guilds, 1000– 1800 (Cambridge: Cambridge University Press, 2011); Sheilagh Ogilvie, The Eu­ro­pean Guilds: An Economic Analy­sis (Prince­ton, NJ: Prince­ton University Press, 2019). 4. Ronald M. Davis and James Rohack, “The AMA & the NMA: Past, Pre­sent, and ­Future” (2008), https://­w ww​.­a ma​-­a ssn​.­org​/­sites​/­default​/­fi les​/­media​-­browser​/­public​/­a ma​ -­history​/­ama​-­nma​-­past​-­present​-­f uture​_­0​.­pdf: https://­w ww​.­ama​-­assn​.­org​/­sites​/­default​/­fi les​ /­media​-­browser​/­public​/­ama​-­history​/­african​-­american​-­physicians3​-­v2​.­pdf. 5. American Bar Association, timelines, http://­w ww​.­a mericanbar​.­org​/­about​_­the​_­aba​ /­timeline​.­html (accessed May 18, 2017). 6. Steven J. Jager, “William Henry Lewis, 1868–1949,” Black Past, July 31, 2012, http://­ www​.­blackpast​.­org​/­aah​/­lewis​-­william​-­henry​-­1868​-­1949. 7. Tad Banicoff, “African Americans and Prince­ton University: A Brief History,” Prince­ ton University Library Guide, March  11, 2005, http://­l ibguides​.­princeton​.­edu​/­c​.­php​?­g​ =­84056&p​=­544526. 8. Douglas S. Massey and Nancy A. Denton, American Apartheid: Segregation and the Making of the Underclass (Cambridge, MA: Harvard University Press, 1993). 9. Kenneth T. Jackson, Crabgrass Frontier: The Suburbanization of the United States (New York: Oxford University Press, 1987). 10. Testimony of Robert D. Atkinson, President, Information Technology and Innovation Foundation, before the House Committee on Oversight and Government Reform Subcommittee on Information Technology, September  26, 2018, http://­w ww2​.­itif​.­org​/­2018​ -­testimony​-­atkinson​-­countering​-­china​.­pdf; Denis Blair and John Huntsman, “Update to the IP Commission Report: The Report of the Commission on the Theft of American Intellectual Property” (Seattle, WA: The National Bureau of Asian Research, February 2017). 11. Victor Nee and Sonja Opper, Capitalism from Below: Markets and Institutional Change in China (Cambridge, MA: Harvard University Press, 2012). 12. J. Fang, H. He, and N. Li, China’s Rising IQ (Innovation Quotient) and Growth: Firm-­ Level Evidence (Washington, DC: International Monetary Fund, 2016). 13. K. Couch and D. Placzek, “Earnings Losses of Displaced Workers Revisited,” American Economic Review 100.1 (2010): 572–89. 14. See historic data from the Job Openings and L ­ abor Turnover Survey from the U.S. Bureau of L ­ abor Statistics. My analy­sis of t­ hese data is in Jonathan T. Rothwell, “Cutting the Losses: Reassessing the Costs of Import Competition to Workers and Communities,” October 19, 2017, https://­ssrn​.­com​/­abstract​=­2920188. 15. Nancy M. Birdsall, Jose Edgardo L. Campos, Chang-­Shik Kim, W. Max Corden, Howard Pack, John Page, Richard Sabor, and Joseph E. Stiglitz, The East Asian Miracle: Economic Growth and Public Policy: Main Report, ed. Lawrence MacDonald, World Bank policy research report (New York: Oxford University Press, 1993), http://­documents​.­worldbank​ .­org​/­curated​/­en​/­975081468244550798​/­Main​-­report. 16. M. Mazzucato, The Entrepreneurial State: Debunking Public vs. Private Sector Myths, vol. 1 (London: Anthem Press, 2015). 17. G. J. Borjas, We Wanted Workers: Unraveling the Immigration Narrative (New York: W. W. Norton, 2016). 18. Raj Chetty, Nathaniel Hendren, Maggie R. Jones, and Sonya R. Porter, Race and Economic Opportunity in the United States: An Intergenerational Perspective, No. w24441

Notes to Chapter 4  •  335 (Cambridge, MA: National Bureau of Economic Research, 2018). See chapter 5 for my discussion of African immigrants to the United States (and Eu­rope), which includes evidence that ­those raised in the United States exceed the education levels of U.S. natives. 19. The econometric model is restricted to the employed U.S.-­born workforce. It regresses the log of total income on binary variables for education, race, and gender. It also includes age, age-­squared, and age-­cubed as controls and year-­fi xed effects, state-­fi xed effects, and broad occupational-­fi xed effects. Errors are clustered at state level. I use a definition of occupations that is consistent over the entire period (which is the occupational variable “occ1990” in the IPUMS database), but for the purposes of occupational-­fi xed effects, I use an aggregation of this into 24 broad occupations. To mea­sure competition from foreign-­born workers, I calculate the percentage of workers who are foreign born by state and occupation. This variable is negative and significantly correlated with the log of income. The coefficient is −0.2. If I use the broad occupational categories instead of the more detailed ones to mea­sure competition from immigration, I get similar results. The data are from Sarah Flood, Miriam King, Renae Rod­gers, Steven Ruggles, and J. Robert Warren, Integrated Public Use Microdata Series, Current Population Survey: Version 6.0 [dataset] (Minneapolis, MN: IPUMS, 2018). https://­ doi​.­org​/­10​.­18128​/­D030​.­V6​.­0. 20. Simonetta Longhi, Peter Nijkamp, and Jacques Poot, “A Meta-­A nalytic Assessment of the Effect of Immigration on Wages,” Journal of Economic Surveys 19.3 (2005): 451–77. 21. Marcus Tullius Cicero, Ethical Writings of Cicero: De Officiis, De Senecture, De Amicitia, and Scorpio’s Dream (Veritatis Splendor Publications, 2014), 498 (Kindle ed.). 22. Cicero, Ethical Writings, 492. 23. Cicero, Ethical Writings, 446. 24. Migration Policy Institute (MPI) tabulation of data from U.S. Census Bureau, 2010– 2017 American Community Surveys (ACS), and 1970, 1990, and 2000 Decennial Census. All other data are from Campbell J. Gibson and Emily Lennon, “Historical Census Statistics on the Foreign-­Born Population of the United States: 1850 to 1990,” Working Paper no. 29 (Washington, DC: U.S. Census Bureau, 1999). 25. Ashley S. Timmer and Jeffrey G. Williamson, “Immigration Policy prior to the Thirties: ­Labor Markets, Policy Interaction, and Globalization Backlash,” Population and Development Review 24.4 (1998): 739–71; P.  H. Lindert and J.  G. Williamson, Unequal Gains: American Growth and In­equality since 1700 (Prince­ton, NJ: Prince­ton University Press, 2016). 26. Immigrant population data from 1850 to 1990 come from the U.S. Census Bureau, https://­w ww​ .­c ensus​ .­g ov​/­p opulation​ /­w ww​/­d ocumentation​ /­t wps0029​ /­t ab01​ .­html, and American Fact Finder for 2000, 2010, and 2017, https://­factfinder​.­census​.­gov​/­faces​/­nav​ /­jsf​/­pages​/­index​.­x html. Birth data are from the National Center for Health Statistics, NCHS Data Visualization Gallery: Natality Trends in the United States, 1909–2015, https://­w ww​.­cdc​.­gov​/­nchs​/­data​-­visualization​/­natality​-­trends​/­index​.­htm. 27. U.S. Census Bureau, American Community Survey, 2017, https://­factfinder​.­census​ .­gov​/­faces​/­nav​/­jsf​/­pages​/­index​.­x html. 28. Douglas S. Massey, Jorge Durand, and Karen A. Pren, “Border Enforcement and Return Migration by Documented and Undocumented Mexicans,” Journal of Ethnic and Migration Studies 41.7 (2015): 1015–40, doi: 10.1080/1369183X.2014.986079. 29. Jonathan Rothwell, “Still Searching: Job Vacancies and STEM Skills” (Washington, DC: The Brookings Institution Metropolitan Policy Program, 2014). 30. Eric Alden Smith, Kim Hill, Frank W. Marlowe, David Nolin, Polly Wiessner, Michael Gurven, Samuel Bowles, Monique Borgerhoff Mulder, Tom Hertz, and Adrian Bell,

336  •  Notes to Chapter 5 “Wealth Transmission and In­equality among Hunter-­Gatherers,” Current Anthropology 51.1 (2010): 19–34. ­These authors estimate a Gini coefficient of in­equality of 0.25 for wealth. That is comparable to Denmark’s Gini coefficient for income and roughly half of the U.S. Gini coefficient for ­house­hold income reported by the Census Bureau for 2017 of 0.48. U.S. wealth in­equality, however, is much more severe, with a Gini coefficient of 0.88, according to Wolff ’s analy­sis of the Survey of Consumer Finances, U.S. Census Bureau, Historical Income ­Tables, “­Table H-4. Gini Indexes for House­holds, by Race and Hispanic Origin of House­holder: 1967 to 2017,” https://­w ww​.­census​.­gov​/­data​/­tables​/­time​-­series​/­demo​/­i ncome​-­poverty​ /­h istorical​-­i ncome​-­i nequality​.­html; Edward N. Wolff, “House­hold Wealth Trends in the United States, 1962 to 2016: Has M ­ iddle Class Wealth Recovered?” No. w24085 (Cambridge, MA: National Bureau of Economic Research, 2017). 5. UNEQUAL ACCESS TO EDUCATION 1. Frederick Douglass, Narrative of the Life of Frederick Douglass [1845] (Word Wise, 2012). 2. Civil Rights Act, 1964, transcript, https://­w ww​.­ourdocuments​.­gov​/­doc​.­php​?­flash​ =­false&doc​=­97&page​=­transcript. 3. James Coleman et al., Equality of Educational Opportunity (Washington, DC: U.S. Department of Health, Wealth, and Education, 1966). 4. Coleman et al., Equality of Educational Opportunity, 21–22. 5. Robert D. Putnam, Our Kids: The American Dream in Crisis (New York: Simon and Schuster, 2016), 182. 6. Gary Ravani, “Why No Excuses Makes No Sense: Revisiting the Coleman Report,” Washington Post, July 23, 2011, https://­w ww​.­washingtonpost​.­com​/­blogs​/­answer​-­sheet​/­post​ /­w hy​-­n o​-­e xcuses​-­m a kes​-­n o​-­s ense​-­r evisiting​-­t he​-­c oleman​-­r eport​/­2 011​/­0 7​/­2 3​ /­gIQAo7W7UI​_­blog​.­html​?­utm​_­term​=­​.­92a1692cb522. 7. Diane Ravitch, “Education: Achievement Gap Starts before School Starts,” mySA, Thursday, October 13, 2011, https://­w ww​.­mysanantonio​.­com​/­community​/­northwest​/­news​ /­article​/­Education​-­Achievement​-­gap​-­starts​-­before​-­school​-­2213710​.­php. 8. Richard J. Herrnstein and Charles Murray, The Bell Curve: Intelligence and Class Structure in American Life (New York: Simon and Schuster, 1994), 394–96. 9. Charles Murray, Coming Apart: The State of White Amer­i­ca, 1960–2010 (New York: Crown Forum, 2012), 60. 10. Murray, Coming Apart, 62. 11. Caroline Hoxby, “The Immensity of the Coleman Data Proj­ect,” Education Next 16.2 (2016). 12. Coleman et al., Equality of Educational Opportunity, ­table 3.221.1, 299. 13. Coleman et al., Equality of Educational Opportunity, ­table 3.221.1, 310. 14. Coleman et al., Equality of Educational Opportunity, ­table 2.41.1, 184. 15. As it happens, he uses many variables to attempt to mea­sure ­family background; t­ hese include parental education, ­whether a f­ ather is pre­sent in the home, parental and student attitudes ­toward education. Together with school effects, t­ hese variables explain 42 ­percent to 53 ­percent of the variation in test scores, which still leaves roughly half of the variation to be explained. In other words, Coleman’s analy­sis shows that schools are the most impor­tant mea­ sured f­ actor in predicting black test scores. It seems a bit strange to then say that schools are of minor importance in his analy­sis.

Notes to Chapter 5  •  337 16. Precisely a one-­standard-­deviation increase in teacher value-­added predicts a 0.27 standard deviation increase in exam scores. A highly effective teacher (at the 75th percentile) can explain a large effect—.33 standard deviations—­relative to a weakly effective teacher (at the 25th percentile). This analy­sis is particularly impressive b­ ecause it controls for test scores at age 14 in the same subject as well as other subjects, thereby allowing the researchers to isolate the effect of subject-­specific teachers. See Helen Slater, Neil M. Davies, and Simon Burgess, “Do Teachers ­Matter? Mea­sur­ing the Variation in Teacher Effectiveness in ­England,” Oxford Bulletin of Economics and Statistics 74.5 (2012): 629–45. 17. Marion F. Shaycoft, “Proj­ect TALENT. The High School Years: Growth in Cognitive Skills. Interim Report 3” (Washington, DC: Office of Education (DHEW), Bureau of Research, 1967). 18. Stuart J. Ritchie and Elliot Tucker-­Drob, “How Much Does Education Improve Intelligence? A Meta-­A nalysis,” Psychological Science 29.8 (2018): 1358–69. 19. I took the average of their reported effect using the raw study-­level data that the authors posted online at https://­osf​.­io​/­r8a24​/­. 20. Shaycoft, “Proj­ect TALENT.” 21. James R. Flynn, “The Mean IQ of Americans: Massive Gains 1932 to 1978,” Psychological Bulletin 95.1 (1984): 29. 22. Jakob Pietschnig and Martin Voracek, “One ­Century of Global IQ Gains: A Formal Meta-­A nalysis of the Flynn Effect (1909–2013),” Perspectives on Psychological Science 10.3 (2015): 282–306. 23. Tamara C. Daley, Shannon E. Whaley, Marian D. Sigman, Michael P. Espinosa, and Charlotte Neumann, “IQ on the Rise: The Flynn Effect in Rural K ­ enyan C ­ hildren,” Psychological Science 14.3 (2003): 215–19. The authors report an increase in IQ of 26.3 using the standard deviation from the 1984 sample. The raw increase corresponds to an increase of 11.2 using standard deviations observed in industrialized countries. 24. Jonathan Rothwell, José Lobo, Deborah Strumsky, and Mark Muro, “Patenting Prosperity: Invention and Economic Per­for­mance in the United States and Its Metropolitan Areas” (Washington, DC: The Brookings Institution, 2016). 25. M. A. Woodley of Menie, M. Peñaherrera-­A guirre, H. B. F. Fernandes, and A.-­J. Figueredo, “What C ­ auses the Anti-­Flynn Effect? A Data Synthesis and Analy­sis of Predictors,” Evolutionary Behavioral Sciences 12.4, 276–95 (forthcoming). 26. Pietschnig and Voracek, “One C ­ entury of Global IQ Gains.” 27. OECD, “Pisa 2015 Results in Focus,” PISA in Focus 67 (2016), https://­w ww​.­oecd​.­org​ /­pisa​/­pisa​-­2015​-­results​-­in​-­focus​.­pdf. 28. Edward Dutton and Richard Lynn, “A Negative Flynn Effect in Finland, 1997–2009,” Intelligence 41.6 (2013): 817–20. 29. Bernt Bratsberg and Ole Rogeberg, “Flynn Effect and Its Reversal Are Both Environmentally Caused,” Proceedings of the National Acad­emy of Sciences 115.26 (2018): 6674–78. 30. Jungho Kim, “Female Education and Its Impact on Fertility,” IZA World of L ­ abor (2016), https://­wol​.­iza​.­org​/­uploads​/­articles​/­228​/­pdfs​/­female​-­education​-­and​-­its​-­impact​-­on​ -­fertility​.­one​-­pager​.­pdf. 31. This analy­sis compared math test scores in 2003 to t­ hose in 2015 using summary data for the country overall and by foreign-­born status via the U.S National Center for Education Statistics International Data Explorer, https://­nces​.­ed​.­gov​/­surveys​/­pisa​/­idepisa/ (accessed June 23, 2018).

338  •  Notes to Chapter 5 32. Raj Chetty, John N. Friedman, and Jonah E. Rockoff, “Mea­sur­i ng the Impacts of Teachers I: Evaluating Bias in Teacher Value-­Added Estimates,” American Economic Review 104.9 (2014): 2593– 2632. 33. Raj Chetty, John N. Friedman, and Jonah E. Rockoff, “Mea­sur­i ng the Impacts of Teachers II: Teacher Value-­Added and Student Outcomes in Adulthood,” American Economic Review 104.9 (2014): 2633–79. 34. Andrew J. Mashburn, Robert C. Pianta, Bridget K. Hamre, Jason T. Downer, Oscar A. Barbarin, Donna Bryant, Margaret Burchinal, Diane M. Early, and Carollee Howes, “Mea­ sures of Classroom Quality in Prekindergarten and ­Children’s Development of Academic, Language, and Social Skills,” Child Development 79.3 (2008): 732–49. 35. Specifically, the difference in math learning between a teacher at the 25th percentile on the CLASS rating and one in the 75th percentile is 0.10 standard deviations for math, 0.05 for language, and 0.23 for student effort. The cognitive effect increases considerably when combined with two other quality measures—­student ratings and the test score gains of prior students—to .24, .12, and .22 for math, language, and student effort, respectively. Student effort was mea­sured by self-­reported willingness to do work outside of the classroom. See Thomas J. Kane and Douglas O. Staiger, “Gathering Feedback for Teaching: Combining High-­Quality Observations with Student Surveys and Achievement Gains,” Mea­sure of Effective Teaching (MET) Proj­ect (Seattle, WA: Bill & Melinda Gates Foundation, 2012). 36. A one-­standard-­deviation increase in the CLASS score caused a .11, .11, and .07 standard deviation increase in math, language, and executive functioning, respectively. See M. Caridad Araujo, Pedro Carneiro, Yyannú Cruz-­Aguayo, and Norbert Schady, “Teacher Quality and Learning Outcomes in Kindergarten,” Quarterly Journal of Economics 131.3 (2016): 1415–53. 37. For my analy­sis, see Jonathan T. Rothwell, “Classroom In­equality and the Cognitive Race Gap: Evidence from 4-­Year-Olds in Public PreK,” March 1, 2016, https://­ssrn​.­com​ /­abstract​=­2740527. For raw data, see D. Early, M. Burchinal, O. Barbarin, D. Bryant, F. Chang, R. Clifford, et al., Pre-­Kindergarten in Eleven States: NCEDL’s Multi-­State Study of Pre-­ Kindergarten and Study of State-­Wide Early Education Programs (SWEEP), ICPSR34877-­v1 (Ann Arbor, MI: Inter-­university Consortium for Po­liti­cal and Social Research [distributor], 2013-10-02), http://­doi​.­org​/­10​.­3886​/­ICPSR34877​.­v1. 38. Raw IQ scores ­were available for the Peabody Picture Vocabulary and Woodcock to Johnson III (math); see Early et al., Pre-­Kindergarten in Eleven States. 39. Black students in top-­quartile classes closed roughly 20 ­percent of the achievement gap on the Peabody vocabulary IQ test when compared to white ­people in top-­quartile classes and 30 ­percent of the gap when compared to white students who did not attend top-­quartile classes. The text refers to a mean of standardized scores across three IQ tests and four learning tests. For related analy­sis, see Jonathan T. Rothwell, “Classroom In­equality and the Cognitive Race Gap: Evidence from 4-­Year-Olds in Public PreK,” (March 1, 2016), https://­ssrn​ .­com​/­abstract​=­2740527 or http://­d x​.­doi​.­org​/­10​.­2139​/­ssrn​.­2740527. 40. Dan Goldhaber, Lesley Lavery, and Roddy Theobald, “Uneven Playing Field? Assessing the Teacher Quality Gap between Advantaged and Disadvantaged Students,” Educational Researcher 44.5 (2015): 293–307. 41. Bridget K. Hamre, Robert C. Pianta, Margaret Burchinal, Samuel Field, Jennifer LoCasale-­Crouch, Jason T. Downer, et al., “A Course on Effective Teacher-­Child Interactions: Effects on Teacher Beliefs, Knowledge, and Observed Practice,” American Educational Research Journal 49.1 (2012): 88–123.

Notes to Chapter 5  •  339 42. H. J. Kitzman, D. L. Olds, R. E. Cole, C. A. Hanks, E. A. Anson, K. J. Arcoleo, et al., “Enduring Effects of Prenatal and Infancy Home Visiting by Nurses on ­Children: Follow-­Up of a Randomized Trial among C ­ hildren at Age 12 Years,” Archives of Pediatrics & Adolescent Medicine 164.5 (2010): 412–18. 43. Timothy J. Bartik, From Preschool to Prosperity: The Economic Payoff to Early Childhood Education (Kalamazoo, MI: Upjohn Institute for Employment Research, 2014), https://­ doi​.­org​/­10​.­17848​/­9780880994835. 44. Greg J. Duncan and Aaron J. Sojourner, “Can Intensive Early Childhood Intervention Programs Eliminate Income-­Based Cognitive and Achievement Gaps?” Journal of ­Human Resources 48.4 (2013): 945–68. 45. Arthur J. Reynolds, Judy A. ­Temple, Dylan L. Robertson, and Emily A. Mann, “Long-­ Term Effects of an Early Childhood Intervention on Educational Achievement and Juvenile Arrest: A 15-­Year Follow-­Up of Low-­Income ­Children in Public Schools,” Journal of the American Medical Association 285.18 (2001): 2339–46. 46. Josh Dawsey, “Trump Derides Protections for Immigrants from ‘Shithole’ Countries,” Washington Post, January  12, 2018, https://­w ww​.­washingtonpost​.­com​/­politics​/­trump​ -­attacks​-­protections​-­for​-­immigrants​-­from​-­shithole​-­countries​-­in​-­oval​-­office​-­meeting​/­2018​ /­01​/­11​/­bfc0725c​-­f 711​-­11e7​-­91af​-­31ac729add94​_ ­story​.­html​?­utm​_­term​=­​.­0ce470468b30. 47. For compelling evidence on the low quality of educational opportunities in poor countries, see UNESCO, “More Than One-­Half of ­Children and Adolescents Are Not Learning Worldwide” (Paris: UNESCO Institute for Statistics, 2017), http://­u is​.­unesco​.­org ​/­sites​ /­default​/­fi les​/­documents​/­fs46 ​-­more​-­than​-­half​-­children​-­not​-­learning​-­en​-­2017​.­pdf; Tessa Bold, Deon Filmer, Gayle Martin, Ezequiel Molina, Brian Stacy, Christophe Rockmore, et al., “Enrollment without Learning: Teacher Effort, Knowledge, and Skill in Primary Schools in Africa,” Journal of Economic Perspectives 31.4 (2017): 185–204; Nazmul Chaudhury, Jeffrey Hammer, Michael Kremer, Karthik Muralidharan, and F. Halsey Rogers, “Missing in Action: Teacher and Health Worker Absence in Developing Countries,” Journal of Economic Perspectives 20.1 (2006): 91–116. 48. Deborah Wilson, Simon Burgess, and Adam Briggs, “The Dynamics of School Attainment of E ­ ngland’s Ethnic Minorities,” Journal of Population Economics 24.2 (2011): 681– 700; see working paper version at http://­eprints​.­lse​.­ac​.­uk​/­6245​/­1​/­The​_­Dynamics​_­of​_ ­School​ _­Attainment​_­of​_­England’s​_­Ethnic​_­M inorities​.­pdf. I used their summary statistics in ­tables 2 and 3 to calculate IQ scores. 49. Feyisa Demie, “Raising the Achievement of Black African Pupils: Good Practice in Schools” (London: Lambeth Research and Statistics Unit, 2013), https://­w ww​.­lambeth​.­gov​ .­u k​/­rsu​/­sites​/­l ambeth​.­g ov​.­u k​.­rsu​/­f iles​/­R aising ​_­t he ​_ ­Achievement​_­of​_ ­Black ​_ ­A frican​ _­Pupils​-­Good​_­Practice​_­in​_ ­Schools​_ ­2013​.­pdf. 50. Demie, “Raising the Achievement of Black African Pupils.” 51. Jenny Easby, “GCSE and Equivalent Attainment by Pupil Characteristics, 2013 to 2014 (Revised),” United Kingdom Department of Education, January 29, 2015, figure 2, https://­w ww​.­gov​.­u k​/­government​/­uploads​/­system​/­uploads​/­attachment​_­data​/­fi le​/­399005​ /­SFR06​_­2015​_­Text​.­pdf. 52. Dominique Lemmermann and Regina T. Riphahn, “The Causal Effect of Age at Migration on Youth Educational Attainment,” Economics of Education Review 63 (2018): 78–99. 53. Lemmermann and Riphahn, “The Causal Effect of Age at Migration on Youth Educational Attainment.”

340  •  Notes to Chapter 5 54. See UNESCO, “More Than One-­Half of ­Children and Adolescents Are Not Learning Worldwide”; Bold et al., “Enrollment without Learning”; Chaudhury et al., “Missing in Action.” For evidence that gray-­matter volume in the frontal lobe of the brain levels off around ages 11 for females and 12 for males, see Jay N. Giedd, Jonathan Blumenthal, Neal O. Jeffries, F. Xavier Castellanos, Hong Liu, Alex Zijdenbos, et al., “Brain Development during Childhood and Adolescence: A Longitudinal MRI Study,” Nature Neuroscience 2.10 (1999): 861. 55. Linda S. Gottfredson, “Mainstream Science on Intelligence: An Editorial with 52 Signatories, History, and Bibliography,” Intelligence 24.1 (1997): 13–23. 56. Robert Plomin and Ian J. Deary, “Ge­ne­tics and Intelligence Differences: Five Special Findings,” Molecular Psychiatry 20.1 (2015): 98. 57. A.  J.  F. Griffiths, Susan  R. Wessler, Richard  C. Lewontin, William  M. Gelbart, David T. Suzuki, and Jeffrey H. Miller, “Quantifying Heritability,” in An Introduction to Ge­ ne­tic Analy­sis, 7th ed. (New York: W. H. Freeman, 2000), https://­w ww​.­ncbi​.­nlm​.­nih​.­gov​ /­books​/­N BK21866​/­. 58. Tinca Polderman, Beben Benyamin, Christiaan A. De Leeuw, Patrick F. ­Sullivan, Arjen Van Bochoven, Peter M. Visscher, and Danielle Posthuma, “Meta-­A nalysis of the Heritability of ­Human Traits Based on Fifty Years of Twin Studies,” Nature Ge­ne­tics 47 (2015): 702–9, http://­match​.­ctglab​.­nl​/­. 59. MaTCH, Meta-­A nalysis of Twin Correlations and Heritability, http://­match​.­ctglab​ .­nl​/­#​/­specific​/­plot1 (accessed June 2, 2019). Despite common claims to the contrary, the heritability estimates do not show a tendency to increase with age. They are lowest for c­ hildren at or below 11 and adults over 65, with no increase from teen years to prime working years. 60. Amelia R. Branigan, Kenneth J. McCallum, and Jeremy Freese, “Variation in the Heritability of Educational Attainment: An International Meta-­A nalysis,” Social Forces 92.1 (2013): 109–40. 61. Hannah Gordon, Frederik Trier Moller, Vibeke Andersen, and Marcus Harbord, “Heritability in Inflammatory Bowel Disease: From the First Twin Study to Genome-­Wide Association Studies,” Inflammatory Bowel Diseases 21.6 (2015): 1428–34. 62. Richard E. Nisbett, Joshua Aronson, Clancy Blair, William Dickens, James Flynn, Diane F. Halpern, and Eric Turkheimer, “Intelligence: New Findings and Theoretical Developments,” American Psychologist 67.2 (2012): 130. 63. Eric Turkheimer, Andreana Haley, Mary Waldron, Brian d’Onofrio, and Irving I. Gottesman, “Socioeconomic Status Modifies Heritability of IQ in Young ­Children,” Psychological Science 14. 6 (2003): 623–28. 64. I am grateful to Eric Turkheimer and Dalton Conley for helping me clarify this over personal communication. 65. Roar Fosse, Jay Joseph, and Ken Richardson, “A Critical Assessment of the Equal-­ Environment Assumption of the Twin Method for Schizo­phre­nia,” Frontiers in Psychiatry 6 (2015): 62. 66. MaTCH, Meta-­A nalysis of Twin Correlations and Heritability, http://­match​.­ctglab​ .­nl​/­#​/­specific​/­plot1 (accessed June 2, 2019). 67. Elizabeth Suhay, Nathan P. Kalmoe, Leon Porter, Perry Silverschanz, Abby Stewart, Laura Stoker, Jennifer Talarico, et al., “The Equal Environment Assumption in Twin Studies of Po­liti­cal Traits: Social Confounds and Suggested Remedies,” unpublished manuscript, 2010, http://­sites​.­lafayette​.­edu​/­suhaye​/­fi les​/­2011​/­01​/ ­Violations​-­ofthe​-­EEA​-­Suhay​-­Kalmoe​ .­pdf.

Notes to Chapter 5  •  341 68. T. J. Bouchard, D. T. Lykken, M. McGue, N. L. Segal, and A. Tellegen, “Sources of ­ uman Psychological Differences: The Minnesota Study of Twins Reared Apart,” Science H 250.4978 (1990): 223–28. The education of a­ dopted m ­ others is strongly similar (0.41) for twins reared apart. The authors suggest that this ­doesn’t ­matter ­because they find no correlation between twin IQ and ­adopted ­mother education, but that is surely a fault of the database’s low sample size (56 twin pairs) and does not reflect the real­ity of the relationship. Adoption studies consistently find that the educational attainment of the a­ dopted ­mother (and f­ ather) strongly predicts IQ. For example, see Mikael Lindahl, Anders Björklund, and Erik Plug, Intergenerational Effects in Sweden: What Can We Learn from Adoption Data? IZA Discussion paper series, No. 1194, 2004. 69. Peter M. Visscher, Sarah E. Medland, Manuel A. R. Ferreira, Katherine I. Morley, Gu Zhu, Belinda K. Cornes, Grant W. Montgomery, et al., “Assumption-­Free Estimation of Heritability from Genome-­Wide Identity-­by-­Descent Sharing between Full Siblings,” PLoS Ge­ ne­tics 2.3 (2006): e41. 70. Alexander I. Young, Michael L. Frigge, Daniel F. Gudbjartsson, Gudmar Thorleifsson, Gyda Bjornsdottir, Patrick Sulem, Gisli Masson, et al., “Estimating Heritability without Environmental Bias,” bioRxiv (Novermber 14, 2017): 218883, https://­w ww​.­biorxiv​.­org​ /­content​/­10​.­1101​/­218883v1. See also Alexander I. Young, Michael L. Frigge, Daniel F. Gudbjartsson, Gudmar Thorleifsson, Patrick Sulem, Gisli Masson, Unnur Thorsteinsdottir, et al., “Relatedness Disequilibrium Regression Estimates Heritability without Environmental Bias,” Nature Ge­ne­tics 50 (2018): 1304–10. 71. Cornelius A. Rietveld, Sarah E. Medland, Jaime Derringer, Jian Yang, Tõnu Esko, Nicolas W. Martin, Harm-­Jan Westra, et al., “GWAS of 126,559 Individuals Identifies Ge­ne­tic Variants Associated with Educational Attainment,” Science 340.6139 (2013): 1467–71. 72. Rietveld et al., “GWAS of 126,559 Individuals Identifies Ge­ne­tic Variants”; for details see supplemental materials at https://­doi​.­org​/­10​.­1126​/­science​.­1235488. 73. Aysu Okbay, Jonathan P. Beauchamp, Mark Alan Fontana, James J. Lee, Tune H. Pers, Cornelius A. Rietveld, Patrick Turley, et al., “Genome-­Wide Association Study Identifies 74 Loci Associated with Educational Attainment,” Nature (London) 533.7604 (2016): 53. 74. Okbay et al., “Genome-­Wide Association Study Identifies 74 Loci”; supplemental information available at http://­ssgac​.­org​/­documents​/­SI​_­74 ​_­loci​_­educational​_­attainment​.­pdf. 75. James J. Lee, Robbee Wedow, Aysu Okbay, Edward Kong, Omeed Maghzian, Megan Zacher, Tuan Ahn Nguyen-­Viet, et al., “Gene Discovery and Polygenic Prediction from a 1.1-­M illion-­Person GWAS of Educational Attainment,” Nature Ge­ne­tics 50.8 (2018): 1112–21. 76. Social Science Ge­ne­tic Association Consortium, Frequently Asked Questions, https://­ www​.­thessgac​.­org ​/­faqs (accessed May 26, 2018). 77. Augustine Kong, Gudmar Thorleifsson, Michael L. Frigge, Bjarni J. Vilhjalmsson, Alexander I. Young, Thorgeir E. Thorgeirsson, Stefania Benonisdottir, et al., “The Nature of Nurture: Effects of Parental Genotypes,” Science 359.6374 (2018): 424–28. 78. D. Belsky, B. Domingue, R. Wedow, L. Arseneault, J. Boardman, A. Caspi, D. Conley, et al., “Ge­ne­tic Analy­sis of Social Class Mobility: Evidence from Five Longitudinal Studies,” Proceedings of the National Acad­emy of Sciences (2018), doi:10.1073/pnas.1801238115. 79. See supplementary text for Lee et al., “Gene Discovery and Polygenic Prediction.” 80. Nicholas Papageorge and Kevin Thom, “Genes, Education, and L ­ abor Market Outcomes: Evidence from the Health and Retirement Study” (May 31, 2017), https://­ssrn​.­com​ /­abstract​=­2982606 or http://­d x​.­doi​.­org​/­10​. ­2139​/­ssrn​. ­2982606.

342  •  Notes to Chapter 6 81. Belsky et al., “Ge­ne­tic Analy­sis of Social Class Mobility.” 82. Belsky et al., “Ge­ne­tic Analy­sis of Social Class Mobility,” found a correlation of 0.28 between parental education and genes for education (mea­sured as the polygenic score) in two samples. This value squared is 8 ­percent, which is the amount of variation explained. They also found a correlation of just 0.12 between “social origins” (parental education, ­house­hold income, and ­father’s occupation) and the polygenic score in their largest sample, which suggests that a more comprehensive mea­sure of class explains only 1 ­percent of the variation in genes that predict education. 83. Plato, The Republic, trans. Benjamin Jowett (Titan Classics, 2015), 1765 (Kindle ed.). 84. Riccardo E. Marioni, Lars Penke, Gail Davies, Jennifer E. Huffman, Caroline Hayward, and Ian J. Deary, “The Total Burden of Rare, Non-­Synonymous Exome Ge­ne­tic Variants Is Not Associated with Childhood or Late-­Life Cognitive Ability,” Proceedings of the Royal Society B 281.1781 (2014): 20140117. “Rare” in this case meant that the allele was pre­sent in less than 1 ­percent of the population ­under study; changing that to 5 ­percent did not change the results. 85. This is based on my analy­sis of supplementary ­table 13 in Lee et al., “Gene Discovery and Polygenic Prediction.” The average SNP that reaches genome-­wide significance has an allele frequency of 0.47. I separately analyzed all ten million SNPs in the authors’ file “GWAS_ CP_all.txt.” The average frequency is the same for all SNPs (.50) as for 13,175 SNPs with p-­values below 5 × 10 −8 (0.5), all of which are common, with allele frequencies above 1 ­percent. The analy­sis ­won’t be applicable to non-­Europeans u­ ntil the polygenic scores can be calculated in larger samples. 86. World Bank data (https://­data​.­worldbank​.­org/ [accessed May 29, 2019]) show that in 2016, 164 countries had per capita income levels that ­were lower than the average for high-­ income countries in 1960, adjusted for inflation, and that many countries had income levels that ­were a tiny fraction of even the 1960 level for rich countries. Thus, one could reasonably expect that IQ scores for ­these countries would be well below the IQ scores of rich countries 50 years ago (15 IQ points). 6. THE HISTORICAL CONTINGENCIES OF GROUP DIFFERENCES IN SKILLS 1. Aristotle, Politics, trans. Benjamin Jowett, book 3, http://­classics​.­mit​.­edu​/­A ristotle​ /­politics​.­3​.­three​.­html. 2. Aristotle, Politics, book 1. 3. Quoted in W. E. B. Du Bois, ed., Black Reconstruction in Amer­i­ca: T ­ oward a History of the Part Which Black Folk Played in the Attempt to Reconstruct Democracy in Amer­i­ca, 1860–1880 (New York: Routledge, 2017), 2000 (Kindle ed.). 4. Nell Irvin Painter, The History of White P ­ eople (New York: W. W. Norton, 2010); Louis Menand, The Metaphysical Club (New York: Macmillan, 2001). 5. Elyce Zenoff Ferster, “Eliminating the Unfit—­Is Sterilization the Answer?” Ohio State Law Journal 27 (1966): 591. 6. “One Hundred and Forty-­Fourth Session,” New York Legislative Documents, 5: 12–43 (Albany, NY: J. B. Lyon Com­pany, 1921). 7. Pearce Baily, Charles Johnson, and Frank Utter, “State Commission for M ­ ental Defectives, Second Annual Report,” in “One Hundred and Forty-­Fourth Session,” New York Legislative Documents, 5: 12–43 (Albany, NY: J. B. Lyon Com­pany, 1921).

Notes to Chapter 6  •  343 8. Thomas C. Leonard, Illiberal Reformers: Race, Eugenics, and American Economics in the Progressive Era (Prince­ton, NJ: Prince­ton University Press, 2016). 9. Ezra Klein, “The Sam Harris Debate,” Vox, April 9, 2018, https://­w ww​.­vox​.­com​/­2018​ /­4​/­9​/­17210248​/­sam​-­harris​-­ezra​-­k lein​-­charles​-­murray​-­transcript​-­podcast. 10. George M. Fredrickson, Racism: A Short History (Prince­ton, NJ: Prince­ton University Press, 2002). 11. Arthur Jensen, “How Much Can We Boost IQ and Scholastic Achievement?” Harvard Educational Review 39.1 (1969): 1–123. 12. Arthur Jensen and J. Phillipe Rushton, “Thirty Years of Research on Race Differences in Cognitive Ability,” Psy­chol­ogy, Public Policy, and Law 11.2 (2005): 235–94. 13. Richard Herrnstein and Charles Murray, The Bell Curve: Intelligence and Class Structure in American Life (New York: ­Free Press: 1994). 14. Linda S. Gottfredson, “Mainstream Science on Intelligence: An Editorial with 52 Signatories, History, and Bibliography,” Intelligence 24.1 (1997) 13–23. 15. Heiner Rindermann, David Becker, and Thomas R. Coyle, “Survey of Expert Opinion on Intelligence: ­Causes of International Differences in Cognitive Ability Tests,” Frontiers in Psy­chol­ogy 7 (2016): 399. 16. Author’s analy­sis of National Assessment of Education Pro­g ress (NAEP) Data Explorer, https://­w ww​.­nationsreportcard​.­gov​/­. Science and reading scores have overlapping 95 ­percent confidence intervals and differed very slightly. 17. National Center for Education Statistics, Digest of Education Statistics, 2015, NCES 2016–014 (Washington, DC: U.S. Department of Education, 2016), chapter 2, https://­nces​ .­ed​.­gov​/­fastfacts​/­display​.­asp​?­id​=­171. 18. Anemona Hartocollis and Stephanie Saul, “Affirmative Action ­Battle Has a New Focus: Asian-­A mericans,” New York Times, August 2, 2017. Economist David Card submitted an analy­sis for the defense in a lawsuit against Harvard, and he concluded that Harvard did not discriminate against Asians b­ ecause Asian candidates scored lower on nonacademic criteria that are valued by the university. See https://­projects​.­iq​.­harvard​.­edu​/­fi les​/­d iverse​ -­education​/­fi les​/­expert​_­report​_­​-­​_ ­2 017​-­12​-­15​_­d r​.­​_­d avid ​_­card ​_­expert​_­report​_­updated​ _­confid​_­desigs​_­redacted​.­pdf. 19. “Jewish Nobel Prize Winners,” JINFO​.­org, http://­w ww​.­jinfo​.­org​/­Nobel​_­Prizes​.­html (accessed May  29, 2019); Charles Murray, “Jewish Genius,” Commentary—­New York–­ American Jewish Committee 123.4 (2007): 29. 20. Gregory Cochran, Jason Hardy, and Henry Harpending, “Natu­ral History of Ashkenazi Intelligence,” Journal of Biosocial Science 38.5 (2006): 1–35; Richard Lynn, “IQ, Socio-­ Economic and Intellectual Achievements of Ashkenazi Jews,” Mankind Quarterly 52.1 (2011): 3–34; Steven Pinker, “Groups and Genes,” New Republic, June 26, 2006; Murray, “Jewish Genius”; Nicholas Wade, A Troublesome Inheritance: Genes, Race and ­Human History (New York: Penguin, 2015). 21. Max Roser and Esteban Ortiz-­Ospina, “Global Rise of Education,” Our World in Data (2019), https://­ourworldindata​.­org​/­g lobal​-­rise​-­of​-­education, using data from the Wittgenstein Centre for Demography and Global H ­ uman Capital (2015), Wittgenstein Centre Data Explorer Version 1.2. 22. Roser and Ortiz-­Ospina, “Global Rise of Education.” 23. In 1937, Jews comprised 3.7 ­percent of the U.S. population and remained above 3 ­percent ­until the late 1960s. Sidney Goldstein, “American Jewry, 1970: A Demographic Profile,” The American Jewish Yearbook 72 (1971), 3–88, http://­w ww​.­jewishdatabank​.­org​ /­Studies​/­downloadFile​.­cfm​?­FileID​=­1447. In 1946, Jews comprised 4.4 ­percent of the U.S.

344  •  Notes to Chapter 6 workforce; “Statistics of Jews,” The American Jewish Yearbook 42 (1940), 589–632, http://­ www​.­ajcarchives​.­org​/­AJC​_ ­DATA​/­Files​/­1940​_­1941​_ ­8 ​_ ­Statistics​.­pdf. 24. Alvin Chenkin, “National Jewish Population Study, Demographic Highlights” (New York: Council of Jewish Federations and Welfare Funds, 1971), https://­w ww​.­jewishdatabank​ .­org ​/­databank​/­search​-­results​/­study​/­304. 25. Maristella Botticini and Zvi Eckstein, The Chosen Few: How Education S­ haped Jewish History, 70-1492 (Prince­ton, NJ: Prince­ton University Press, 2012). 26. Botticini and Eckstein, The Chosen Few. 27. William V. Harris, Ancient Literacy (Cambridge, MA: Harvard University Press, 1989). 28. Harris, Ancient Literacy. 29. Botticini and Eckstein, The Chosen Few, 72. 30. Murray, “Jewish Genius”; Wade, A Troublesome Inheritance. 31. Michael Goldfarb, Emancipation: How Liberating Eu­rope’s Jews from the Ghetto Led to Revolution and Re­nais­sance (New York: Simon and Schuster, 2009). 32. Lynn, “IQ, Socio-­Economic and Intellectual Achievements of Ashkenazi Jews.” 33. Herman Rosenthal, J. G. Lipman, Vasili Rosenthal, L. Wygodsky, M. Mysh, and Abraham Galante, “Rus­sia,” Jewish Encyclopedia, 1906, http://­w ww​.­jewishencyclopedia​.­com​ /­articles​/­12943​-­russia. 34. Rosenthal et al., “Rus­sia.” 35. “Statistics of the Jews,” The American Jewish Year Book, 20 (1918–1919): 339–352, http://­w ww​.­ajcarchives​.­org​/­AJC​_ ­DATA​/­Files​/­1918​_­1919​_­7​_ ­Statistics​.­pdf. 36. Bernard Wasserstein, On the Eve: The Jews of Eu­rope before the Second World War (New York: Simon and Schuster, 2012). 37. Joseph Jacobs and M. Grunwald, “Occupations,” The Jewish Encyclopedia, 1906, http://­w ww​.­jewishencyclopedia​.­com​/­articles​/­11652​-­occupations. 38. Fabian Waldinger, “Quality M ­ atters: The Expulsion of Professors and the Consequences for PhD Student Outcomes in Nazi Germany,” Journal of Po­liti­cal Economy 118.4 (2010): 787–831; Petra Moser, Alessandra Voena, and Fabian Waldinger, “German Jewish Émigrés and US Invention,” American Economic Review 104.10 (2014): 3222–55. 39. Paul Windolf, “The German-­Jewish Economic Elite (1900 to 1930),” Zeitschrift für Unternehmensgeschichte 56.2 (2011): 135–62. 40. Harry S. Linfield, “The Communal Organ­ization of the Jews in the United States, 1927,” The American Jewish Year Book 31 (1929): 99–254. 41. Kaufmann Kohler and Joseph Jacobs, “Intermarriage,” The Jewish Encyclopedia, 1906, http://­w ww​.­jewishencyclopedia​.­com​/­articles​/­8137​-­intermarriage. 42. Petra Moser, “Taste-­Based Discrimination at the NYSE—­Empirical Evidence from a Shock to Preferences a­ fter WWI” (August 20, 2010), https://­ssrn​.­com​/­abstract​=­930237 or http://­d x​.­doi​.­org​/­10​.­2139​/­ssrn​.­930237. 43. Linfield, “The Communal Organ­ization of the Jews.” 44. Windolf, “The German-­Jewish Economic Elite.” 45. I estimate that ­there ­were 1.8 million Episcopalians and 2.3 million Presbyterians living in the United States in 1926, based on data from the 1926 Census of Religious Bodies, from the Association of Religious Data Archives, http://­w ww​.­thearda​.­com​/­A rchive​/­Files​ /­Downloads​/­1936CENSCT​_ ­DL2​.­asp. ­There ­were 4.2 million Jews according to Harry S. Linfield, “The Communal Organ­ization of the Jews in the United States, 1927,” The American Jewish Year Book 31 (1929): 99–254.

Notes to Chapter 6  •  345 46. Moser, Voena, and Waldinger, “German Jewish Émigrés.” 47. Ross Thomson, Structures of Change in the Mechanical Age: Technological Innovation in the United States, 1790–1865 (Baltimore, MD: Johns Hopkins University Press, 2009); B. Zorina Khan, The Democ­ratization of Invention: Patents and Copyrights in American Economic Development, 1790–1920 (Cambridge: Cambridge University Press, 2005). 48. Joel Mokyr, “The Economics of Being Jewish,” Critical Review 23.1–2 (2011): 195–206. 49. “Willem Einthoven,” Wikipedia, The ­Free Encyclopedia, https://­en​.­wikipedia​.­org​/­w​ /­index​.­php​?­title​=­Willem​_ ­Einthoven&oldid​=­895209158 (accessed May 19, 2019). 50. Robin Walker, African American Contributions to Science and Technology, Reklaw Education Lecture Series Book 12 (Reklaw Education, 2012). 51. “Lewis Howard Latimer,” Wikipedia, The ­Free Encyclopedia, https://­en​.­wikipedia​.­org​ /­w​/­index​.­php​?­title​=­Lewis​_ ­Howard​_ ­Latimer&oldid​=­897120552 (accessed May 19, 2019); National Inventor’s Hall of Fame, https://­w ww​.­invent​.­org​/­inductees​/­lewis​-­latimer. 52. National Inventor’s Hall of Fame, https://­w ww​.­invent​.­org​/­inductees​/­elijah​-­mccoy. 53. “George Washington Carver,”  Wikipedia, The F ­ ree Encyclopedia,  https://­en​ .­wikipedia​.­org​/­w​/­index​.­php​?­title​=­George​_­Washington​_­Carver&oldid​=­893757901  (accessed May 19, 2019). 54. National Inventor’s Hall of Fame, https://­w ww​.­i nvent​.­org ​/­i nductees​/­george​ -­washington​-­carver. 55. “Sarah Boone,” Wikipedia, The F ­ ree Encyclopedia, https://­en​.­wikipedia​.­org​/­w​/­index​ .­php​?­title​=­Sarah ​_ ­Boone&oldid​=­893843422 (accessed May  19, 2019); “Miriam Benjamin,”  Wikipedia, The ­ Free Encyclopedia,  https://­en​.­wikipedia​.­org​/­w​/­index​.­php​?­title​ =­Miriam​_­Benjamin&oldid​=­887264204 (accessed May 19, 2019). 56. Henry E. Baker, “The Negro in the Field of Invention,” Journal of Negro History 2.1 (1917): 21–36. See also U.S. Patent and Trademark Office, “African Americans Hold Patents on Impor­tant Inventions,” February 9, 2002, https://­w ww​.­uspto​.­gov​/­about​-­us​/­news​ -­updates​/­african​-­americans​-­hold​-­patents​-­important​-­inventions. 57. “Ellen Eglin,” Wikipedia, The F ­ ree Encyclopedia, https://­en​.­wikipedia​.­org​/­w​/­index​ .­php​?­title​=­Ellen​_ ­Eglin&oldid​=­889447040 (accessed May 19, 2019). 58. The 1.8 ­percent figure is from Ufuk Akcigit, John Grigsby, and Tom Nicholas, The Rise of American Ingenuity: Innovation and Inventors of the Golden Age, no. w23047 (Cambridge, MA: National Bureau of Economic Research, 2017). The 3–10 ­percent figure is from Sarada, Michael Andrews, and Nicola Ziebarth, “Historical Changes in the Demographics of Inventors in the United States” (January 30, 2017), https://­ssrn​.­com​/­abstract​=­2908160. 59. I used full-­count 1920 census data from IPUMS USA to estimate black population shares in the South and outside of it. Steven Ruggles, Sarah Flood, Ronald Goeken, Josiah Grover, Erin Meyer, Jose Pacas, and Matthew Sobek. IPUMS USA: Version 9.0 [1920] (Minneapolis, MN: IPUMS, 2019), https://­doi​.­org​/­10​.­18128​/­D010​.­V9​.­0. 60. Lisa D. Cook, “Inventing Social Capital: Evidence from African American Inventors, 1843–1930,” Explorations in Economic History 48.4 (2011): 507–18. 61. Akcigit, Grigsby, and Nicholas, The Rise of American Ingenuity, figure 4B. 62. Akcigit, Grigsby, and Nicholas, “Immigration and the Rise of American Ingenuity,” 327–31. 63. Baker, “The Negro in the Field of Invention.” 64. Solomon Northrup and David Wilson, “Twelve Years a Slave: Narrative of Solomon Northrup, a Citizen of New-­York, Kidnapped in Washington City in 1841, and Rescued in

346  •  Notes to Chapter 6 1853, from a Cotton Plantation Near the Red River, Louisiana,” in I Was Born a Slave: An Anthology of Classic Slave Narratives, ed. Yuval Taylor and Marshall Joseph Becker, repr. in African Diaspora Archaeology Newsletter 6.3 (1999): article 7, https://­scholarworks​.­umass​.­edu​ /­adan​/­vol6​/­iss3​/­7. 65. Christianna Elrene and Thomas Hurford, “ ‘In His Arm the Scar’: Medicine, Race, and the Social Implications of the 1721 Inoculation Controversy on Boston,” PhD dissertation, Ohio State University, 2010. 66. J. McKeen Cattell, “A Further Statistical Study of American Men of Science,” Science 32.827 (1910): 633–48; J. McKeen Cattell, “A Statistical Study of American Men of Science III. The Distribution of American Men of Science,” Science 24.623 (1906): 732–42. 67. Frederick Douglass, “At Last, at Last, the Black Man Has a ­Future: An Address Delivered in Albany, NY, on 22 April 1870,” Albany Eve­ning Journal, April 23, 1970, http://­ frederickdouglass​.­infoset​.­io​/­islandora​/­object​/­islandora%3A2429. 68. “Madam C. J. Walker,” Wikipedia, The ­Free Encyclopedia, https://­en​.­wikipedia​.­org​ /­w​/­index​.­php​?­title​=­Madam​_­C ​.­​_ ­J​.­​_­Walker&oldid​=­896974633 (accessed May 19, 2019). 69. Lynn, “IQ, Socio-­Economic and Intellectual Achievements of Ashkenazi Jews.” 70. Lynn, “IQ, Socio-­Economic and Intellectual Achievements of Ashkenazi Jews.” 71. Richard Lynn, The Chosen P ­ eople: A Study of Jewish Intelligence and Achievement (Augusta, GA: Washington Summit Publishers, 2011). 72. Lynn, The Chosen P ­ eople. 73. J. Philippe Rushton, “Race, Intelligence, and the Brain: The Errors and Omissions of the ‘Revised’ Edition of S. J. Gould’s The Mismea­sure of Man (1996),” Personality and Individual Differences 23.1 (1997): 169–80; Lynn, “IQ, Socio-­Economic and Intellectual Achievements of Ashkenazi Jews”; Murray, “Jewish Genius.” 74. Audrey  M. Shuey, The Testing of Negro Intelligence (New York: Social Science Press, 1966). 75. Data from the 1936 Census of Religious Bodies is from the Association of Religious Data Archives, http://­w ww​.­thearda​.­com​/­A rchive​/­Files​/­Downloads​/­1936CENSCT​_ ­DL2​ .­asp. 76. This figure is from the 1930 U.S. Census, using data from IPUMS NHGIS (Steven Manson, Jonathan Schroeder, David Van Riper, and Steven Ruggles. IPUMS National Historical Geographic Information System: Version 13.0 [Database] (Minneapolis: University of Minnesota, 2018), http://­doi​.­org​/­10​.­18128​/­D050​.­V13​.­0. 77. Erich Rosenthal, “The Equivalence of United States Census Data for Persons of Rus­sian Stock or Descent with American Jews: An Evaluation,” Demography 12.2 (1975): 275–90. 78. Carl C. Brigham, A Study of American Intelligence (Prince­ton, NJ: Prince­ton University Press, 1923), t­ able 28. 79. Charles S. Berry, “The Classification by Tests of Intelligence of Ten Thousand First-­ Grade Pupils,” Journal of Educational Research 6.3 (1922): 185–203. 80. U. S. Bureau of Labor Statistics, National Longitudinal Surveys, https://www.bls​.gov​ /nls/home.htm. 81. For a cautious analy­sis ignored by Lynn of the scores of Anglo-­Saxons in New York compared to Jews, see Zecharia Dershowitz, Yaakov Frankel, and Brendan A. Maher, “Jewish Culture and the WISC and WAIS Test Patterns,” Journal of Consulting and Clinical Psy­ chol­ogy 43.2 (1975): 126–34; Z. Dershowitz, “Influences of Cultural Patterns on the Thinking of ­Children in Certain Ethnic Groups: A Study of the Effect of Jewish Subculture on the Field-­Dependence-­Independence Dimension of Cognition,” unpublished doctoral dis-

Notes to Chapter 6  •  347 sertation, New York University, 1966. Dershowitz’s results provide evidence that Jewish ­children had the same IQ scores as Anglo-­Saxon ­children with the same class background. 82. Lauress L. Wise and Donald Hatch McLaughlin, The Proj­ect TALENT Data Bank Handbook (Palo Alto, CA: American Institutes for Research, 1977), https://­w ww​ .­projecttalent​.­org​/­about​/­history​/­. The IQ score is labeled “IQ composite” in the database and consists of 25 ­percent abstract reasoning, 24 ­percent mathematical reasoning, and 51 ­percent reading comprehension. 83. Samson D. Oppenheim, “The Jewish Population of the United States,” The American Jewish Year Book 20 (1918), 31–74. 84. Berry, “The Classification by Tests of Intelligence.” 85. Yiddish Melbourne, Monash University, http://­f uture ​.­a rts​.­monash​.­edu​/­y iddish​ -­melbourne​/­statistics/ (accessed February 20, 2018). 86. The entire U.S. population was estimated to be 3 ­percent Jewish at the time, according to World Jewish Population, http://­w ww​.­ajcarchives​.­org​/­AJC​_ ­DATA​/­Files​/­1961​_­16​ _­W JP​.­pdf. 87. In other ways, the Proj­ect Talent data appear to be representative of the Jewish population. Among the c­ hildren of Yiddish speakers in the database, 28 ­percent have at least one parent with a bachelor’s degree, compared with 19 ­percent of c­ hildren from non-­Yiddish-­ speaking ­house­holds. Likewise, 36  ­percent of the ­fathers of Yiddish-­speaking ­children worked in a job as a professional, man­ag­er, or business owner, compared to 21 ­percent of non-­ Yiddish-­speaking white ­fathers. Boris Levinson, “A Comparison of the Per­for­mance of Bilingual and Monolingual Native-­Born Jewish Preschool ­Children of Traditional Parentage on Four Intelligence Tests,” Journal of Clinical Psy­chol­ogy 15.1 (1959): 74–76, finds that the ­fathers of applicants to private school in New York City had a similar share working in professional and managerial roles (37 ­percent). 88. Levinson, “A Comparison of the Per­for­mance of Bilingual and Monolingual Native-­ Born Jewish Preschool ­Children.” I adjusted the WISC scores by 1 IQ point to adjust for the Flynn effect, since Levinson used the 1949 standard on a 1959 population. 89. James Coleman et al., Equality of Educational Opportunity (Washington, DC: U.S. Department of Health, Wealth, and Education: 1966), ­table 9. Asian first-­g rade students scored higher than white ­people on nonverbal, but all other scores shown suggest lower Asian scores. The mean is 50 and standard deviation is 10, according to Coleman. Translating this into an IQ score with mean 100 and standard deviation of 15 is straightforward. 90. Thomas Sowell, “Ethnicity and IQ,” American Spectator 28 (February 1, 1995), 32; Thomas Sowell, Ethnic Amer­i­ca: A History (New York: Basic Books, 1981). 91. Chinese National Statistics, “[­Table] 2–14 Population Aged 6 and Over by Sex, Educational Attainment and Region” (2015), http://­w ww​.­stats​.­gov​.­cn​/­tjsj​/­ndsj​/­2016​/­indexeh​.­htm. 92. Tom Loveless, “PISA’s China Prob­lem” (Washington, DC: The Brown Center Chalkboard, The Brookings Institution, 2013); Duoduo Xu and Jaap Dronkers, “Mi­grant ­Children in Shanghai: A Research Note on the PISA-­Shanghai Controversy,” Chinese So­ cio­log­i­cal Review 48.3 (2016): 271–95. 93. Amanda Wise, “­Behind Singapore’s PISA Rankings Success and Why Other Countries May Not Want to Join the Race,” The Conversation, December 8, 2016, http://­w ww​.­abc​ .­net​.­au​/­news​/­2016​-­12​-­08​/­behind​-­singapore​-­pisa​-­rankings​-­success​/­8104074. 94. Singapore, Ministry of Education, Education Statistics Digest (2017), xiv, https://­w ww​ .­moe​.­gov​.­sg​/­docs​/­default​-­source​/­document​/­publications​/­education​-­statistics​-­d igest​/­esd​ _­2017​.­pdf.

348  •  Notes to Chapter 6 95. Survey of Adult Skills, OECD Skills Surveys, 2011–2015, https://­w ww​.­oecd​.­org​ /­skills​/­piaac​/­data​/­. 96. Knut Tveit, “The Development of Popu­lar Literacy in the Nordic Countries. A Comparative Historical Study,” Scandinavian Journal of Educational Research 35.4 (1991): 241–52. 97. Samuel Bowles and Herbert Gintis, A Cooperative Species: H ­ uman Reciprocity and Its Evolution (Prince­ton, NJ: Prince­ton University Press, 2011), 449–50. 98. Deirdre N. McCloskey, Bourgeois Dignity: Why Economics ­Can’t Explain the Modern World (Chicago: University of Chicago Press, 2010). 99. Adam Fox, “Words, Words, Words: Education and Literacy and Print,” in A Social History of E ­ ngland, 1500–1750, ed. Keith Wrightson (Cambridge: Cambridge University Press, 2017), 134. 100. J. S. Reynolds, “Virtue and Intelligence u­ nder Foot: Reconstruction in South Carolina,” in Walter Lynwood Fleming, Documentary History of Reconstruction, Po­liti­cal, Military, Social, Religious, Educational & Industrial, 1865 to the Pre­sent Time, ed. Walter Lynwood Fleming (Cleveland, OH: A. H. Clark Com­pany, 1907), 1: 424–25. 101. Du Bois, Black Reconstruction in Amer­i­ca. 102. Eric Foner, Reconstruction: Amer­i­ca’s Unfinished Revolution, 1863–1877 (New York: Harper Collins, 2011). 103. Du Bois, Black Reconstruction in Amer­i­ca, 15149 (Kindle ed.). 104. Walter Lynwood Fleming, Documentary History of Reconstruction, Po­liti­cal, Military, Social, Religious, Educational & Industrial, 1865 to the Pre­sent Time, vol. 2 (Cleveland, OH: A.H. Clark Com­pany, 1907), 184–85. 105. IPUMS-­USA, Version 9.0 [Datasets for 1870, 1880, 1890, 1900, 1910]. Minneapolis, MN: IPUMS, 2019, https://doi.org/10.18128/D010.V9.0; restricted to population with at least one U.S.-born parent between ages 20 and 30. 106. Trevon D. Logan, Do Black Politicians ­Matter? no. w24190 (Cambridge, MA: National Bureau of Economic Research, 2018). 107. Nicole Spear, “­Adopted C ­ hildren’s Outcomes as Young Adults in Regards to Educational Attainment and Income,” Park Place Economist 17 (2009), http://­digitalcommons​ .­iwu​.­edu​/­parkplace​/­vol17​/­iss1​/­16; Anders Björklund, Mikael Lindahl, and Erik Plug, “The Origins of Intergenerational Associations: Lessons from Swedish Adoption Data,” Quarterly Journal of Economics 121.3 (2006): 999–1028. 108. Data analyzed are from the U.S. Bureau of Labor Statistics, National Longitudinal Survey of Youth (NLSY) 1997 cohort, 1997–2013 (rounds 1–16), https://www.bls.gov/nls​ /nlsy97.htm. We can also use this for the 1979 version: U.S. Bureau of Labor Statistics, National Longitudinal Survey of Youth (NLSY) 1979 cohort, 1979–2012 (rounds 1–25), https://www.bls.gov/nls/nlsy79.htm. 109. Loren M. Marulis and Susan B. Neuman, “The Effects of Vocabulary Intervention on Young ­Children’s Word Learning: A Meta-­A nalysis,” Review of Educational Research 80.3 (2010): 300–35. 110. David K. Dickinson, Julie A. Griffith, Roberta Michnick Golinkoff, and Kathy Hirsh-­Pasek, “How Reading Books Fosters Language Development around the World,” Child Development Research (2012): 1–15, https://­w ww​.­h indawi​.­com​/­journals​/­cdr​ /­2 012​/­6 02807​/­​.­ Eileen  T. Rodriguez, Catherine  S. Tamis-­L eMonda, Mark  E. Spellman, Barbara A. Pan, Helen Raikes, Julieta Lugo-­Gil, and Gayle Luze, “The Formative Role of Home Literacy Experiences across the First Three Years of Life in C ­ hildren from

Notes to Chapter 6  •  349 Low-­I ncome Families,” Journal of Applied Developmental Psy­ chol­ ogy 30.6 (2009): 677–94. 111. John S. Hutton, Tzipi Horowitz-­K raus, Alan L. Mendelsohn, Tom DeWitt, Scott K. Holland, and the C-­M IND Authorship Consortium, “Home Reading Environment and Brain Activation in Preschool C ­ hildren Listening to Stories,” Pediatrics 136.3 (2015): 466–78. 112. Benjamin N. York, Susanna Loeb, and Christopher Doss, “One Step at a Time: The Effects of an Early Literacy Text Messaging Program for Parents of Preschoolers,” Journal of ­Human Resources (2018): 0517-8756R, http://­jhr​.­uwpress​.­org​/­content​/­early​/­2018​/­01​/­03​/­jhr​ .­54​.­3​.­0517​-­8756R; Matthew A. Kraft and Manuel Monti-­Nussbaum, “Can Schools Enable Parents to Prevent Summer Learning Loss? A Text-­Messaging Field Experiment to Promote Literacy Skills,” Annals of the American Acad­emy of Po­liti­cal and Social Science 674.1 (2017): 85–112. 113. Dickinson et al., “How Reading Books Fosters Language Development.” 114. Frank Ridzi, Monica Sylvia, Xiaofen Qiao, and Jeff Craig, “The Imagination Library Program and Kindergarten Readiness: Evaluating the Impact of Monthly Book Distribution,” Journal of Applied Social Science 11.1 (2017): 11–24. 115. Shahin Samiei, Andrew J. Bush, Marie Sell, and Doug Imig, “Examining the Association between the Imagination Library Early Childhood Literacy Program and Kindergarten Readiness,” Reading Psy­chol­ogy 37.4 (2016): 601–26. The authors report a standard deviation effect of 0.43 on literacy tests, which translates into 6.45 IQ points. The effect was also significant but not as large for math (0.27 standard deviations). 116. ­These data are from the National Longitudinal Survey of Youth, 1979, https://­w ww​ .­bls​.­gov​/­nls​/­nlsy79​.­htm. 117. Amy Hsin and Yu Xie, “Explaining Asian Americans’ Academic Advantage over Whites,” Proceedings of the National Acad­emy of Sciences 111.23 (2014): 8416–21. 118. Feyisa Demie and Christabel McLean, “Raising the Achievement of African-­ Heritage Pupils: A Case Study of Good Practice in British Schools,” Educational Studies 33.4 (2007): 415–34; Osagie Festus Amayo, “Dimensions of Challenging Parenting Practices: Nigerian Immigrants in the United States,” Journal of Pan-­African Studies 3.2 (2009): 83–100. 119. David H. Uttal, “Beliefs about Ge­ne­tic Influences on Mathe­matics Achievement: A Cross-­Cultural Comparison,” Ge­ne­tica 99.2–3 (1997): 165–72. 120. Uttal, “Beliefs about Ge­ne­tic Influences on Mathe­matics Achievement,” 171. 121. E. Bettinger, S. Ludvigsen, M. Rege, I. F. Solli, and D. Yeager, “Increasing Perseverance in Math: Evidence from a Field Experiment in Norway,” Journal of Economic Be­hav­ior & Organ­ization 146 (2018): 1–15. 122. Feyisa Demie and Kirstin Lewis, “White Working Class Achievement: An Ethnographic Study of Barriers to Learning in Schools,” Educational Studies 37.3 (2011): 245–64. 123. “About Dr Anne-­Marie Imafidon MBE,” http://­aimafidon​.­com​/­about​/­. 124. “Won­der Twins: Seven-­Year-­Olds Are Youn­gest Ever to Pass A-­Level Maths,” Daily Mail, January 15, 2009, http://­w ww​.­dailymail​.­co​.­u k​/­news​/­article​-­1116584​/­Wonder​-­t wins​ -­Seven​-­year​-­olds​-­youngest​-­pass​-­A S​-­level​-­maths​.­html#ixzz5B4FoArs6; http://­ireport​.­cnn​ .­com​/­docs​/­DOC​-­317180. 125. Laura Roberts and John Bingham, “GCSE Results: Girl, Five, Claims Rec­ord While Seven-­Year-­Old Boy Receives A*,” Daily Telegraph, August 25, 2010, https://­w ww​.­telegraph​

350  •  Notes to Chapter 7 .­co​.­u k​/­education​/­educationnews​/­7962546​/­G CSE​-­results​-­Girl​-­five​-­claims​-­record​-­while​ -­seven​-­year​-­old​-­boy​-­receives​-­A ​.­html. 126. “A-­Level Success for Twins, Eight,” BBC News, August 20, 2009, http://­news​.­bbc​.­co​ .­u k​/­2​/­hi​/­u k​_­news​/­england​/­berkshire​/­8211160​.­stm (accessed May 29, 2019). 7. UNEQUAL ACCESS TO HOUSING MARKETS 1. Andre M. Perry, Jonathan Rothwell, and David Harshbarger, “The Devaluation of Assets in Black Neighborhoods: The Case of Residential Property” (Washington, DC: The Brookings Institution, 2018). 2. David M. Cutler, Edward L. Glaeser, and Jacob L. Vigdor, “The Rise and Decline of the American Ghetto,” Journal of Po­liti­cal Economy 107.3 (1999): 455–506. 3. Douglas S. Massey and Nancy A. Denton, American Apartheid: Segregation and the Making of the Underclass (Cambridge, MA: Harvard University Press, 1993); Ta-­Nehisi Coates, “The Case for Reparations,” Atlantic, June 2014; Richard Rothstein, The Color of Law: A Forgotten History of How Our Government Segregated Amer­i­ca (New York: Liveright, 2017). 4. William Easterly, “Empirics of Strategic Interdependence: The Case of the Racial Tipping Point,” The BE Journal of Macroeconomics 9.1 (2009), https://­w ww​.­degruyter​.­com​/­view​ /­j​/­bejm​. ­2009​.­9​.­1​/­bejm​. ­2009​.­9​.­1​.­1492​/­bejm​. ­2009​.­9​.­1​.­1492​.­xml. 5. Jonathan T. Rothwell, “Racial Enclaves and Density Zoning: The Institutionalized Segregation of Racial Minorities in the United States,” American Law and Economics Review 13.1 (2011): 290–358. 6. Data from 1890–1960 are from Cutler, Glaeser, and Vigdor, “The Rise and Decline of the American Ghetto.” Jacob Vigdor graciously shared data with me. Their 1890–1930 data rely on census wards, which are local voting districts; from 1940–2010, all estimates are based on census tracts, which are like neighborhoods. Wards are much larger and therefore artificially more encompassing, suggesting less segregation. For example, in the District of Columbia, in 2010, ­there ­were only 8 wards, with populations ranging from 70,000 to 80,000, but t­ here w ­ ere 179 census tracts with populations ranging from just u­ nder 1,000 to 7,500. To make the ward data comparable, I follow Douglas S. Massey, Jonathan Rothwell, and Thurston Domina, “The Changing Bases of Segregation in the United States,” Annals of the American Acad­emy of Po­liti­cal and Social Science 626.1 (2009): 74–90, and regress tract segregation mea­sures on ward segregation mea­sures for 1940, when both are available. The resulting equation is used to adjust ward data upward by multiplying by a coefficient (.5 for dissimilarity and .96 for isolation) and adding a constant term (.436 and .166, respectively). Data from 1980–2010 are from John Logan, available at https://­s4​.­ad​.­brown​.­edu​ /­projects​/­diversity​/­Data​/­Download1​.­htm; see also John R. Logan and Brian J. Stults, “The Per­sis­tence of Segregation in the Metropolis: New Findings from the 2010 Census,” census brief prepared for Proj­ ect US2010 (2011), https://s4.ad.brown.edu/Projects/Diversity​ /Data/Report/report2.pdf. Data for 1970 w ­ ere calculated by the author using Steven Manson, Jonathan Schroeder, David Van Riper, and Steven Ruggles, IPUMS National Historical Geographic Information System [NHGIS]: Version 12.0 [Database] (Minneapolis: University of Minnesota Press, 2017), http://­doi​.­org​/­10​.­18128​/­D050​.­V12​.­0. 7. Author analy­sis of IPUMS NHGIS, www​.­nhgis​.­org. 8. U.S. Department of the Treasury, Internal Revenue Service, “SOI Tax Stats—Individual Income Tax Statistics—ZIP Code Data,” https://www.irs.gov/statistics/soi-tax-stats​-indi​ vidual-income-tax-statistics-zip-code-data-soi (accessed June 19, 2019).

Notes to Chapter 7  •  351 9. Gallup-Sharecare Well-Being Index, 2016–2017, https://wellbeingindex.sharecare.com/. 10. Louis Menand, The Metaphysical Club (London: Macmillan, 2001); Nell Irvin Painter, The History of White P ­ eople (New York: W. W. Norton, 2010). 11. “WAR IN ‘HELL’S KITCHEN’: Negroes and Whites Renew Rioting in Eighth Ave­nue and Many Are Hurt,” New York Times, August 4, 1898. 12. “WAR IN ‘HELL’S KITCHEN.’ ” 13. Cutler, Glaeser, and Vigdor, “The Rise and Decline of the American Ghetto.” 14. Thomas Lee Philpott, The Slum and the Ghetto (New York: Oxford University Press, 1978), 159. 15. Philpott, The Slum and the Ghetto. 16. Philpott, The Slum and the Ghetto, 169. 17. Dale Cockrell, “Jim Crow, Demon of Disorder,” American ­Music 14.2 (1996): 161–84. 18. Ira Katznelson, When Affirmative Action Was White: An Untold History of Racial In­ equality in Twentieth-­Century Amer­i­ca (New York: W. W. Norton, 2005). 19. “Whites Object to P ­ eople of Color in the Third Ward,” Chicago Defender, September 22, 1917, 1. 20. Philpott, The Slum and the Ghetto, 165. 21. Gunnar Myrdal, An American Dilemma: The Negro Prob­lem and Modern Democracy (New York: Harper B ­ rothers Publishers, 1944), 82. 22. Sam Stavinsky, “500 Attend Rally to Prevent Sale of Homes to Negroes: 500 Protest Home Sales to Negroes,” Washington Post, November 9, 1947, M1. 23. Cutler, Glaeser, and Vigdor, “The Rise and Decline of the American Ghetto.” 24. Rothstein, The Color of Law. 25. Philpott, The Slum and the Ghetto, 353. 26. Rothstein, The Color of Law. 27. Rothstein, The Color of Law; Swedish economist Gunnar Myrdal also emphasized the importance of covenants in this influential study. Gunnar Myrdal, An American Dilemma: The Negro Prob­lem and Modern Democracy (New York: Harper B ­ rothers, 1944). 28. Massey and Denton, American Apartheid; Arnold R. Hirsch, Making the Second Ghetto: Race and Housing in Chicago, 1940–1960 (Chicago: University of Chicago Press, 2009); Rothstein, The Color of Law. 29. Massey and Denton, American Apartheid. 30. Margery Austin Turner, Robert Santos, Diane  K. Levy, Douglas  A. Wissoker, Claudia Aranda, Robb Pitingolo, “Housing Discrimination against Racial and Ethnic Minorities, 2012. Washington, DC: US Department of Housing and Urban Development,” Policy Development and Research, June 11, 2013, https://­w ww​.­urban​.­org​/­research​ /­p ublication​/­housing​-­d iscrimination​-­a gainst​-­racial​-­a nd​-­e thnic​-­m inorities​-­2 012​-­f ull​ -­report. 31. Jacob S. Rugh and Douglas S. Massey, “Racial Segregation and the American Foreclosure Crisis,” American So­cio­log­i­cal Review 75.5 (2010): 629–51. 32. Marvin M. Smith and Christy Chung Hevener, “Subprime Lending over Time: The Role of Race,” Journal of Economics and Finance 38.2 (2014): 321–44. 33. Henry N. Pontell, William K. Black, and Gilbert Geis, “Too Big to Fail, Too Power­ ful to Jail? On the Absence of Criminal Prosecutions ­a fter the 2008 Financial Meltdown,” Crime, Law and Social Change 61.1 (2014): 1–13; see also William K. Black discussion with Russ Roberts, “William Black on Financial Fraud,” Library of Economics and Liberty, February 6, 2010, http://­w ww​.­econtalk​.­org​/­archives​/­2012​/­02​/­william​_­black ​_­o​.­html.

352  •  Notes to Chapter 7 34. Binyamin Applebaum, “A Nonprofit Lender Revives the Hopes of Subprime Borrowers,” New York Times, February 25, 2014. 35. William A. Fischel, “Zoning Rules” (Cambridge, MA: Lincoln Institute of Land Policy, 2015); Edward L. Glaeser and Bryce A. Ward, “The ­Causes and Consequences of Land Use Regulation: Evidence from Greater Boston,” Journal of Urban Economics 65.3 (2009): 265–78; Edward Glaeser, Triumph of the City: How Urban Spaces Make Us H ­ uman (London: Pan Macmillan, 2011). 36. Jonathan T. Rothwell, “Racial Enclaves and Density Zoning: The Institutionalized Segregation of Racial Minorities in the United States,” American Law and Economics Review 13.1 (2011): 290–358; Jonathan T. Rothwell and Douglas S. Massey, “Density Zoning and Class Segregation in US Metropolitan Areas,” Social Science Quarterly 91.5 (2010): 1123–43; Jonathan Rothwell and Douglas S. Massey, “The Effect of Density Zoning on Racial Segregation in US Urban Areas,” Urban Affairs Review 44.6 (2009): 779–806; Jonathan Rothwell, Housing Costs, Zoning, and Access to High-­Scoring Schools (Washington, DC: The Brookings Institution, 2012). 37. Jason Rhodes, “The Value of Exclusion: Chasing Scarcity through Social Exclusion in Early Twentieth-­Century Atlanta,” ­Human Geography 9.1 (2016): 46–67. 38. David M. P. Freund, Colored Property: State Policy and White Racial Politics in Suburban Amer­i­ca (Chicago: University of Chicago Press, 2010). 39. Thomas C. Leonard, “American Economic Reform in the Progressive Era: Its Foundational Beliefs and Their Relation to Eugenics,” History of Po­liti­cal Economy 41.1 (2009): 109–41. 40. Rhodes, “The Value of Exclusion,” 57. 41. Quoted in Freund, Colored Property, 66. 42. Ruth Knack, Stuart Meck, and Israel Stollman, “The Real Story b­ ehind the Standard Planning and Zoning Acts of the 1920s,” Land Use Law & Zoning Digest 48.2 (1996): 3–9. 43. Village of Euclid v. Ambler Realty Co., 272 U.S. 365 (1926). 44. United States National Commission on Urban Prob­lems, Building the American City—­Report of the National Commission on Urban Prob­lems to the Congress and to the President of the United States (Washington, DC, 1968). 45. Marie Boyd, “Zoning for Apartments: A Study of the Role of Law in the Control of Apartment Houses in New Haven, Connecticut, 1912–1932,” Pace Law Review 33 (2013): 600. 46. Boyd, “Zoning for Apartments.” 47. Boyd, “Zoning for Apartments.” 48. Fischel, “Zoning Rules.” 49. K. Einstein, M. Palmer, and D. Glick, “Who Participates in Local Government? Evidence from Meeting Minutes,” Perspectives on Politics 17.1 (2018): 1–19. 50. Jessica Trounstine, “The Geography of In­equality: How Land Use Regulation Produces Segregation and Polarization,” University of California Merced Working Paper, July 2018. 51. Rothwell and Massey, “Density Zoning and Class Segregation.” 52. Kendra Bischoff and Sean  F. Reardon, “Residential Segregation by Income, ­1970–2009,” in Diversity and Disparities: Amer­i­ca Enters a New C ­ entury, ed. John Logan (New York: Russell Sage, 2014), 208–33. 53. Rothwell and Massey, “Density Zoning and Class Segregation.” 54. Gallup World Poll, https://www.gallup.com/analytics/232838/world-poll.aspx (accessed June 19, 2019).

Notes to Chapter 7  •  353 55. The detailed wording is as follows: “Would you recommend the city or area where you live to a friend or associate as a place to live, or not?” “Are you satisfied or dissatisfied with the city or area where you live?” “Do you feel safe walking alone at night in the city or area where you live?” Fi­nally, three questions use the same preface: “In the city or area where you live, are you satisfied or dissatisfied with (1) the educational system or the schools? (2) the availability of good affordable housing? (3) the availability of quality healthcare?” 56. Nada Issa, “Tunisia’s Dirty Secret: Five Years a­ fter the Revolution, Tunisia’s Black Minority Has Yet to Experience the Freedoms Enjoyed by Other Citizens,” Al Jazeera, March 17, 2016, http://­w ww​.­a ljazeera​.­com​/­programmes​/­peopleandpower​/­2016​/­03​/­tunisia​ -­dirty​-­secret​-­160316153815980​.­html. 57. Siiri Silm and Rein Ahas, “The Temporal Variation of Ethnic Segregation in a City: Evidence from a Mobile Phone Use Dataset,” Social Science Research 47 (2014): 30–43. 58. David Beckett, “Trends in the United Kingdom Housing Market, 2014” (London: UK Office of National Statistics, 2014), http://­webarchive ​.­nationalarchives ​.­gov​.­u k​ /­20160109144125​/­http://­w ww​.­ons​.­gov​.­uk​/­ons​/­dcp171766​_­373513​.­pdf. U.S. data are from the U.S. Agency for Housing and Urban Development, “American Housing Survey for the United States: 2001” (Washington, DC: 2002), https://­w ww​.­census​.­gov​/­prod​/­2002pubs​ /­h150​-­01​.­pdf. 59. John Iceland, Pablo Mateos, and Gregory Sharp, “Ethnic Residential Segregation by Nativity in ­Great Britain and the United States,” Journal of Urban Affairs 33.4 (2011): 409–29. 60. Jonathan Laurence and Justin Vaisse, “Understanding Urban Riots in France” (Washington, DC: The Brookings Institution, December 1, 2005), https://­w ww​.­brookings​ .­edu​/­articles​/­understanding​-­urban​-­riots​-­in​-­france​/­. 61. Tanvi Misra, “The Othered Paris,” City Lab, November 16, 2016, https://­w ww​.­citylab​ .­com​/­equity​/­2017​/­11​/­the​-­othered​-­paris​/­543597​/­. 62. British data are from Ceri Peach, “Slippery Segregation: Discovering or Manufacturing Ghettos?” Journal of Ethnic and Migration Studies 35.9 (2009): 1381–95. U.S. data are from analy­sis using IPUMS National Historical Geographic Information System, Version 12.0 (Minneapolis: University of Minnesota, 2017). 63. I use the K = 6,400 threshold, which mea­sures the number of neighbors used in the neighborhood; of ­those published this is closest to the U.S. census tract median of 4,000 residents. Smaller definitions also yield smaller segregation mea­sures. See B. Malmberg, M. M. Nielsen, E. Andersson, and K. Haandrikman, “Residential Segregation of Eu­ro­pean and Non-­European Mi­g rants in Sweden: 1990–2012,” Eu­ro­pean Journal of Population 34.2 (2018): 169–93. 64. Sako Musterd, “Social and Ethnic Segregation in Eu­rope: Levels, C ­ auses, and Effects,” Journal of Urban Affairs 27.3 (2005): 331–48. 65. Sako Musterd, Szymon Marcińczak, Maarten Van Ham, and Tjit Tammaru, “Socioeconomic Segregation in Eu­ro­pean Capital Cities. Increasing Separation between Poor and Rich,” Urban Geography 38.7 (2017): 1062–83. 66. Matthew A. Light, “Dif­fer­ent Ideas of the City: Origins of Metropolitan Land-­Use Regimes in the United States, Germany, and Switzerland,” Yale Journal of International Law 24 (1999): 577–611; Sonia Hirt, “Mixed Use by Default: How the Eu­ro­pe­ans (­Don’t) Zone,” Journal of Planning Lit­er­a­ture 27.4 (2012): 375–93. 67. Hirt, “Mixed Use by Default.” 68. Light, “Dif­fer­ent Ideas of the City,” 594.

354  •  Notes to Chapter 8 69. Light, “Dif­fer­ent Ideas of the City,” 595. Emphasis in original. 70. Sonia Hirt, “Home, Sweet Home: American Residential Zoning in Comparative Perspective,” Journal of Planning Education and Research 33.3 (2013): 292–309; Hirt, “Mixed Use by Default.” 8. HOW UNEQUAL ACCESS TO HOUSING PERPETUATES GROUP IN­E QUALIT Y AND INJUSTICE 1. Jessica Trounstine, “Segregation and In­equality in Public Goods,” American Journal of Po­liti­cal Science 60.3 (2016): 709–25. 2. R. Paternoster, J. M. McGloin, H. Nguyen, and K. J. Thomas, “The Causal Impact of Exposure to Deviant Peers: An Experimental Investigation,” Journal of Research in Crime and Delinquency 50.4 (2013): 476–503. 3. Anne C. Case and Lawrence F. Katz, The Com­pany You Keep: The Effects of F ­ amily and Neighborhood on Disadvantaged Youths, no. w3705 (Cambridge, MA: National Bureau of Economic Research, 1991). 4. Geoffrey T. Wodtke, David J. Harding, and Felix Elwert, “Neighborhood Effects in Temporal Perspective: The Impact of Long-­Term Exposure to Concentrated Disadvantage on High School Graduation,” American So­cio­log­i­cal Review 76.5 (2011): 713–36; Patrick Sharkey and Felix Elwert, “The Legacy of Disadvantage: Multigenerational Neighborhood Effects on Cognitive Ability,” American Journal of Sociology 116.6 (2011): 1934–81; Geoffrey T. Wodtke, “Duration and Timing of Exposure to Neighborhood Poverty and the Risk of Adolescent Parenthood,” Demography 50.5 (2013): 1765–88. 5. Jonathan Rothwell and Douglas Massey, “Geographic Effects on Intergenerational Income Mobility,” Economic Geography 91.1 (2015): 83–106. 6. Raj Chetty, John N. Friedman, Nathaniel Hendren, Maggie R. Jones, and Sonya R. Porter, The Opportunity Atlas: Mapping the Childhood Roots of Social Mobility, no. w25147 (Cambridge, MA: National Bureau of Economic Research, 2018), see figure 3, “Geographic Decomposition of Variance in Upward Mobility.” 7. Raj Chetty, Nathaniel Hendren, Maggie R. Jones, and Sonya R. Porter, Race and Economic Opportunity in the United States: An Intergenerational Perspective, no. w24441 (Cambridge, MA: National Bureau of Economic Research, 2018). 8. Raj Chetty and Nathaniel Hendren, “The Impacts of Neighborhoods on Intergenerational Mobility I: Childhood Exposure Effects,” Quarterly Journal of Economics 133.3 (2018): 1107–62; Raj Chetty and Nathaniel Hendren, “The Impacts of Neighborhoods on Intergenerational Mobility II: County-Level Estimates,” Quarterly Journal of Economics 133.3 (2018): 1163–1228. 9. O. Åslund, P.-­A . Edin, P. Fredriksson, and H. Gronqvist, “Peers, Neighborhoods, and Immigrant Student Achievement: Evidence from a Placement Policy,” American Economic Journal: Applied Economics 3.2 (2011): 67–95. 10. P.-A. Edin, P. Fredriksson, and O. Åslund, “Ethnic Enclaves and the Economic Success of Immigrants—­Evidence from a Natu­ral Experiment,” Quarterly Journal of Economics 118.1 (2003): 329–57. 11. Eric D. Gould, Victor Lavy, and M. Daniele Paserman, “Sixty Years ­a fter the Magic Carpet Ride: The Long-­Run Effect of the Early Childhood Environment on Social and Economic Outcomes,” Review of Economic Studies 78.3 (2011): 938–73.

Notes to Chapter 8  •  355 12. E. D. Gould, V. Lavy, and M. D. Paserman, “Immigrating to Opportunity: Estimating the Effect of School Quality Using a Natu­ral Experiment on Ethiopians in Israel,” Quarterly Journal of Economics 119.2 (2004): 489–526. 13. Jonathan Rothwell, “Sociology’s Revenge: Moving to Opportunity (MTO) Revisited,” The Brookings Institution, Social Mobility Memos, May 6, 2015, https://­w ww​.­brookings​.­edu​ /­blog ​/­social​-­mobility​-­memos​/­2 015​/­05​/­0 6​/­sociologys​-­revenge​-­moving​-­to​-­opportunity​ -­mto​-­revisited​/­. 14. Rothwell, “Sociology’s Revenge.” 15. Jeffrey R. Kling, Jeffrey B. Liebman, and Lawrence F. Katz, “Experimental Analy­sis of Neighborhood Effects,” Econometrica 75.1 (2007): 108. 16. J. Ludwig, G. J. Duncan, L. A. Gennetian, L. F. Katz, R. C. Kessler, J. R. Kling, and L. Sanbonmatsu, “Neighborhood Effects on the Long-­Term Well-­B eing of Low-­I ncome Adults,” Science 337.6101 (2012): 1509. 17. Eleonora Patacchini and Yves Zenou, “Neighborhood Effects and Parental Involvement in the Intergenerational Transmission of Education,” Journal of Regional Science 51.5 (2011): 987–1013. 18. Rebecca Casciano and Douglas S. Massey, “School Context and Educational Outcomes: Results from a Quasi-­Experimental Study,” Urban Affairs Review 48.2 (2012): 180– 204; https://­ssrn​.­com​/­abstract​=­1865232 or http://­d x​.­doi​.­org​/­10​. ­2139​/­ssrn​.­1865232. 19. Heather Schwartz, “Housing Policy Is School Policy: Eco­nom­ically Integrative Housing Promotes Academic Success in Montgomery County, MD,” Education Digest 76.6 (2011): 42. 20. Sven Hernberg, “Lead Poisoning in a Historical Perspective,” American Journal of Industrial Medicine 38.3 (2000): 244–54; Herbert Needleman, “Lead Poisoning,” Annual Review of Medicine 55 (2004): 209–22. 21. Joel Schwartz, “Low-­Level Lead Exposure and ­Children’s IQ: A Meta-­A nalysis and Search for a Threshold,” Environmental Research 65.1 (1994): 42–55; J. Liu, L. Li, Y. Wang, C. Yan, and X. Liu, “Impact of Low Blood Lead Concentrations on IQ and School Per­for­ mance in Chinese C ­ hildren,” PLoS One 8.5 (2013): e65230. 22. Bruce P. Lanphear, Richard Hornung, Jane Khoury, Kimberly Yolton, Peter Baghurst, David C. Bellinger, Richard L. Canfield, et al., “Low-­Level Environmental Lead Exposure and C ­ hildren’s Intellectual Function: An International Pooled Analy­sis,” Environmental Health Perspectives 113.7 (2005): 894–99. 23. David C. Bellinger and Andrew M. Bellinger, “Childhood Lead Poisoning: The Torturous Path from Science to Policy,” Journal of Clinical Investigation 116.4 (2006): 853–57; David Rosner and Gerald Markowitz, “Building the World That Kills Us: The Politics of Lead, Science, and Polluted Homes, 1970 to 2000,” Journal of Urban History 42.2 (2016): 323–45; Richard Rabin, “Warnings Unheeded: A History of Child Lead Poisoning,” American Journal of Public Health 79.12 (1989): 1668–74. 24. Centers for Disease Control and Prevention (CDC), “Blood Lead Levels in C ­ hildren Aged 1–5 Years—­United States, 1999–2010,” MMWR: Morbidity and Mortality Weekly Report 62.13 (2013): 245. 25. Richard G. Newell and Kristian Rogers, “The US Experience with the Phasedown of Lead in Gasoline” (Washington, DC: Resources for the ­Future, 2003). 26. Bellinger and Bellinger, “Childhood Lead Poisoning.” 27. This is also true to some extent for Hispanics. Ingrid Gould Ellen, David M. Cutler, and William Dickens, “Is Segregation Bad for Your Health? The Case of Low Birth Weight [with comments],” Brookings-­Wharton Papers on Urban Affairs (2000): 203–38.

356  •  Notes to Chapter 8 28. Douglas S. Massey, “Why Death Haunts Black Lives,” Proceedings of the National Acad­emy of Sciences 114.5 (2017): 800–802. 29. Rebecca Casciano and Douglas S. Massey, “Neighborhood Disorder and Anxiety Symptoms: New Evidence from a Quasi-­Experimental Study,” Health & Place 18.2 (2012): 180–90. 30. Gallup. For description, see “How Does the Gallup U.S. Daily Work?,” https://­w ww​ .­gallup​.­com​/­174146​/­gallup​-­daily​-­methodology​.­aspx (accessed May 23, 2019). 31. See Opportunity Insights, https://­opportunityinsights​.­org​/­(accessed May 23, 2019). Chetty et  al., The Opportunity Atlas; Jonathan Rothwell, “The Biggest Economic Divides ­Aren’t Regional. ­They’re Local. (Just Ask Parents.),” New York Times, February 12, 2019. 32. Rachel E. Morgan and Grace Kena, “Criminal Victimization, 2016: Revised,” U.S. Department of Justice, Bureau of Justice Statistics, NCJ 251150 (October 2018), https://­w ww​ .­bjs​.­gov​/­content​/­pub​/­pdf​/­c v16​.­pdf. For hom­i­cide data, see the Federal Bureau of Investigation, Crime in the USA, 2015, Uniform Crime Report, expanded t­ able 6, https://­ucr​.­fbi​.­gov​ /­crime​-­in​-­the​-­u​.­s​/­2015​/­crime​-­in​-­the​-­u​.­s​.­​-­2015​/­tables​/­expanded ​_­homicide​_­data ​_­table​_ ­6​ _­murder​_­race ​_ ­and ​_ ­sex ​_­of​_­vicitm​_­by ​_­race ​_ ­and ​_ ­sex ​_­of​_­offender​_ ­2015​.­x ls. 33. David J. De Wit, Ellen Lipman, Maria Manzano-­Munguia, Jeffrey Bisanz, Kathryn Graham, David R. Offord, Elizabeth O’Neill, et al., “Feasibility of a Randomized Controlled Trial for Evaluating the Effectiveness of the Big B ­ rothers Big S­ isters Community Match Program at the National Level,” ­Children and Youth Ser­vices Review 29.3 (2007): 383–404. 34. District of Columbia Public Schools, Shaw M ­ iddle School, 2012–2013 School Scorecard, http://­profiles​.­dcps​.­dc​.­gov​/­pdf​/­shaw2012​.­pdf (accessed May 23, 2019). 35. Theola Labbé-­DeBose, “Mourners Gather a­ fter Fatal Shooting of a Promising 15-­Year-­Old Boy,” Washington Post, June 3, 2011. 36. Lizette Alvarez and Cara Buckley, “Zimmerman Is Acquitted in Killing of Trayvon Martin,” New York Times, July 14, 2013, http://­w ww​.­nytimes​.­com​/­2013​/­07​/­15​/­us​/­george​ -­zimmerman​-­verdict​-­trayvon​-­martin​.­html. 37. Jasmine C. Lee and Haeyoun Park, “15 Black Lives Ended in Confrontations with Police. 3 Officers Convicted,” New York Times, October 5, 2018, https://­w ww​.­nytimes​.­com​ /­interactive​/­2017​/­05​/­17​/­us​/­black​-­deaths​-­police​.­html. 38. Thomas F. Pettigrew and Linda R. Tropp, When Groups Meet: The Dynamics of Intergroup Contact (London: Psy­chol­ogy Press, 2013). 39. Jonathan T. Rothwell, “The Effects of Racial Segregation on Trust and Volunteering in US Cities,” Urban Studies 49.10 (2012): 2109–36. 40. Roland G. Fryer Jr., Reconciling Results on Racial Differences in Police Shootings, no. w24238 (Cambridge, MA: National Bureau of Economic Research, 2018). 41. Roland G. Fryer Jr., “An Empirical Analy­sis of Racial Differences in Police Use of Force,” Journal of Po­liti­cal Economy 127.3 (2019), https://­w ww​.­journals​.­uchicago​.­edu​/­doi​ /­f ull​/­10​.­1086​/­701423. 42. Cody T. Ross, Bruce Winterhalder, and Richard McElreath, “Resolution of Apparent Paradoxes in the Race-­Specific Frequency of Use-­of-­Force by Police,” Palgrave Communications 4.1 (2018): 61. 43. New York Civil Liberties Union, “Stop-­and-­Frisk-­Data,” https://­w ww​.­nyclu​.­org​/­en​ /­stop​-­and​-­Frisk​-­data (accessed May 23, 2019). 44. Demographic data are from New York City Department of City Planning, http://­ maps​.­nyc​.­gov​/­census​/­, using the precinct office address as the neighborhood; search data are

Notes to Chapter 8  •  357 from author analy­sis of raw 2014 data from Stop-­and-­Frisk Data, released by the New York City Police Department and posted online by the New York City Civil Liberties Union, https://­w ww​.­nyclu​.­org​/­en​/­stop​-­and​-­frisk​-­data. 45. Philip Bump, “The Facts about Stop-­and-­Frisk in New York City,” Washington Post, September 26, 2016. 46. Author analy­sis of data from Howard N. Snyder, Alexia D. Cooper, and Joseph Mulako-­Wangota, “Arrests in the United States, 1980–2014,” FBI, Uniform Crime Reporting Program, Bureau of Justice Statistics, August 2017, https://­w ww​.­bjs​.­gov​/­index​.­cfm​?­t y=­datool​ &surl​=­​/­arrests​/­index​.­cfm#. 47. U.S. Department of Health and ­Human Ser­vices, Substance Abuse and ­Mental Health Ser­vices Administration, Center for Behavioral Health Statistics and Quality, National Survey on Drug Use and Health, 2013, ICPSR35509-­v3 (Ann Arbor, MI: Inter-­University Consortium for Po­liti­cal and Social Research [distributor], 2015), https://­doi​.­org​/­10​.­3886​ /­ICPSR35509​.­v3. 48. One can perform this calculation (the probability of being arrested conditional on selling drugs) directly using the database, but it would be hopelessly biased downward, ­because ­people who ­were arrested and then incarcerated, even briefly, could not have completed the survey. Thus, the survey can capture only p­ eople who ­were arrested and not incarcerated or incarcerated e­ arlier in the year and l­ater freed. It works better as a mea­sure of p­ eople who “got away with it.” 49. I used a logit regression in STATA to perform this calculation. T-­stat on black binary variable is 4.7. Sample size is 6,662 men aged 18 to 30 who ­were e­ ither black or white. If I drop age restrictions, I estimate that black males are 4.5 times more likely to be arrested on drug charges, with a z-­statistic of 6.3 (sample size is 18,670). U.S. Department of Health and ­Human Ser­vices, National Survey on Drug Use and Health, 2013. I defined common illegal drugs as marijuana, heroin, crack, cocaine, LSD, and meth. Drug sales question asked: “During the past 12 months, how many times have you sold illegal drugs?” 50. Author analy­sis of raw 2014 data from Stop-­and-­Frisk Data, https://­w ww​.­nyclu​.­org​ /­en​/­stop​-­and​-­frisk​-­data. 51. Centers for Disease Control and Prevention, National Center for Health Statistics, Compressed Mortality File 1999–2016 on CDC WON­DER Online Database, released December 2017. Data are from the Compressed Mortality File 1999–2016, series 20, no.  2V, 2017, http://­wonder​.­cdc​.­gov​/­cmf​-­icd10 ​. ­html, as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program. My analy­sis is of deaths from two major ICD-10 categories: “F10–­F19 (­Mental and behavioral disorders due to psychoactive substance use),” which include opioids, cocaine, marijuana, tobacco; and “X40–­X49 (Accidental poisoning by and exposure to noxious substances).” 52. Federal Bureau of Investigation, Crime in the United States, 2013, https://­ucr​.­fbi​.­gov​ /­crime​-­in​-­the​-­u​.­s​/­2013​/­crime​-­in​-­the​-­u​.­s​.­​-­2013​/­tables​/­table​- ­43. 53. United States Department of Health and ­Human Ser­v ices, Substance Abuse and ­Mental Health Ser­vices Administration, Center for Behavioral Health Statistics and Quality, National Survey on Drug Use and Health, 2013, ICPSR35509-­v1 (Ann Arbor, MI: Inter-­ University Consortium for Po­liti­cal and Social Research [distributor], 2014-11-18), http://­ doi​.­org​/­10​.­3886​/­ICPSR35509​.­v1. 54. Federal Bureau of Investigation, Crime in the United States, 2013.

358  •  Notes to Chapter 9 55. For thorough discussion of this topic, see Michelle Alexander, The New Jim Crow: Mass Incarceration in the Age of Colorblindness (New York: New Press, 2010). 56. Jonathan Rothwell, “Drug Offenders in American Prisons: The Critical Distinction between Stock and Flow” (Washington, DC: The Brookings Institution, November 25, 2015). 57. Bruce Western, “The Impact of Incarceration on Wage Mobility and In­equality,” American So­cio­log­i­cal Review 67.4 (2002): 526–46; Isabel V. Sawhill, Scott Winship, and Kerry Searle Grannis, “Pathways to the M ­ iddle Class: Balancing Personal and Public Responsibilities,” Issues in Science and Technology 29.2 (2013): 47–54. 9. UNEQUAL ACCESS TO THE BUYING AND SELLING OF PROFESSIONAL SER­V ICES 1. Darryl R. Biggar and Michael Owen Wise, “Competition in Professional Ser­vices” (February  3, 2000), OECD, Best Practice Roundtables in Competition Policy No.  27, https://­ssrn​.­com​/­abstract​=­318763 or http://­d x​.­doi​.­org​/­10​. ­2139​/­ssrn​.­318763. 2. Sheilagh Ogilvie, “The Economics of Guilds,” Journal of Economic Perspectives 28.4 (2014): 169–92. 3. Author analy­sis of data from OECD.STAT, “Health Care Resources: Remuneration of Health Professionals,” https://­stats​.­oecd​.­org ​/­(accessed April 3, 2018). 4. My analy­sis of IPUMS USA 2012–2016 American Community Survey, using the “occsoc” category “physicians and surgeons”; see Steven Ruggles, Sarah Flood, Ronald Goeken, Josiah Grover, Erin Meyer, Jose Pacas, and Matthew Sobek, IPUMS USA: Version 9.0 [2012– 2016 American Community Survey] (Minneapolis, MN: IPUMS, 2019), https://­doi​.­org​/­10​ .­18128​/­D010​.­V9​.­0​.­ 5. Eurostat data show that 37 ­percent of doctors in Belgium are general prac­ti­tion­ers and that all doctors are compensated at a level 4.5 times that of the average Belgian; physicians, by specialty, http://­ec​.­europa​.­eu​/­eurostat​/­statistics​-­explained​/­index​.­php​?­title​=­File:​ Physicians,​_­by ​_­speciality,​_ ­2015​_­H LTH17​.­png (accessed May  30, 2019). The corresponding figure for the United States using IPUMS USA data from 2012–2016 is 4.9, as reported in the text. 6. My analy­sis of IPUMS USA, 2012–2016 data from the U.S. Census Bureau’s American Community Survey. 7. “Is Medical School Worth It Financially?,” Best Medical Degrees, http://­w ww​ .­bestmedicaldegrees​.­com​/­is​-­medical​-­school​-­worth​-­it​-­fi nancially​/­(accessed June 4, 2019); “Net Price at For-­Profit Institutions by De­pen­dency Status and Income, 2011–12,” Trends in Higher Education, https://­trends​.­collegeboard​.­org​/­college​-­pricing​/­figures​-­tables​/­net​-­price​ -­for​-­profit​-­institutions​-­dependency​-­status​-­income​-­2011​-­12 (accessed June 4, 2019). 8. Leiyu Shi and Douglas A. Singh, “The Evolution of Health Ser­v ices in the United States,” chap. 3 in Delivering Health Care in America: A Systems Approach (Burlington, MA: Jones & Bartlett Learning, 2014). 9. Institute of Medicine, The F ­ uture of Nursing: Leading Change, Advancing Health (Washington, DC: The National Academies Press, 2011). 10. M. Laurant, D. Reeves, R. Hermens, J. Braspenning, R. Grol, and B. Sibbald, “Substitution of Doctors by Nurses in Primary Care,” Cochrane Database of Systematic Reviews no. 4 (2004): art. no.: CD001271, doi: 10.1002/14651858.CD001271.pub2.

Notes to Chapter 9  •  359 11. Chris Hafner-­Eaton and Laurie K. Pearce, “Birth Choices, the Law, and Medicine: Balancing Individual Freedoms and Protection of the Public’s Health,” Journal of Health Politics, Policy and Law 19.4 (1994): 813–35. 12. Kenneth C. Johnson and Betty-­A nne Daviss, “Outcomes of Planned Home Births with Certified Professional Midwives: Large Prospective Study in North Amer­i­ca,” BMJ 330.7505 (2005): 1416. 13. Brian Dulisse and Jerry ­Cromwell, “No Harm Found When Nurse Anesthetists Work without Supervision by Physicians,” Health Affairs 29.8 (2010): 1469–75. 14. American Association of Nurse Prac­ti­tion­ers, “State Practice Environment,” https://­ www​.­a anp​.­org​/­legislation​-­regulation​/­state​-­legislation​/­state​-­practice​-­environment​/­6 6​ -­legislation​-­regulation​/­state​-­practice​-­environment​/­1380​-­state​-­practice​-­by​-­t ype (accessed September 2017). 15. American College of Nurse-­Midwives, State Resource Center, http://­w ww​.­midwife​ .­org ​/­State​-­Resource​-­Center (accessed May 30, 2019); American Association of Nurse Anesthetists, “State Legislative and Regulatory Requirements (50-­state summaries),” http://­w ww​ .­a ana​.­com​/­a dvocacy​/­stategovernmentaffairs​/­Pages​/­Federal​-­Supervision​-­R ule​-­Opt​-­O ut​ -­Information​.­aspx (accessed May 30, 2019). 16. Stephen Isaacs and Paul Jellinek, “Accept No Substitute: A Report on Scope of Practice” (The Physicians Foundation, November 2012), 1. 17. Isaacs and Jellinek, “Accept No Substitute,” 1. 18. Isaacs and Jellinek, “Accept No Substitute.” 19. Isaacs and Jellinek, “Accept No Substitute.” 20. Data in this paragraph are from FollowtheMoney​.­org, https://­w ww​.­followthemoney​ .­org​/­(accessed May 23, 2019). 21. California Legislative Information, SB-323 Nurse Prac­ti­tion­ers: Scope of Practice, https://­leginfo​.­legislature​.­ca​.­gov​/­faces​/­billVotesClient​.­xhtml​?­bill​_­id​=­201520160SB323 (accessed May 23, 2019); contributions data are from FollowtheMoney​.­org, https://­w ww​.­follow​ themoney​.­org​/­(accessed May 23, 2019). 22. Elizabeth Aguilera, “Nurse Practitioner Autonomy Bill Fails in State Assembly (updated),” June 30, 2015, http://­w ww​.­scpr​.­org ​/­news​/­2015​/­0 6​/­30​/­52800​/­nurse​-­practitioner​ -­autonomy​-­bill​-­fails​-­in​-­state​-­as​/­. 23. ­These data are from FollowtheMoney​.­org, which compiles state campaign disclosure rec­ords. The funding ­here applies to 2014 and 2016 campaigns; see https://­w ww​.­follow​ themoney​.­org​/­(accessed May 23, 2019). 24. California Medical Association, “Dangerous Bill Putting Patients in Harm’s Way Barely Passes Assembly Committee,” Press Release, August 13, 2013, http://­w ww​.­cmanet​ .­org​/­news​/­press​-­detail​/­​?­article​=­dangerous​-­bill​-­putting​-­patients​-­in​-­harms​-­way. 25. The American Medical Association, House of Delegates, 2017, https://­w ww​.­ama​-­assn​ .­org ​/­sites​/­default​/­fi les​/­media​-­browser​/­public​/­hod​/­i17​-­refcommb​-­a nnotated​.­pdf (accessed April 13, 2018); see congratulatory press release supporting the resolution against in­de­pen­ dent nurse practice from the American Society of Anesthesiologists, https://­w ww​.­asahq​.­org​ /­a dvocacy​/­fda​-­a nd​-­washington​-­a lerts​/­washington​-­a lerts​/­2 017​/­11​/­a ma​-­a dopts​-­a sa​-­led​ -­resolution​-­opposing​-­the​-­onerous​-­aprn​-­initiative. 26. The American Medical Association, “Scope of Practice,” https://­w ww​.­ama​-­assn​.­org​ /­about​/­scope​-­practice (accessed April 13, 2018). See also American Medical Association, “Physician Assistant Scope of Practice,” Advocacy Resource Center, https://­w ww​.­ama​-­assn​.­org​/­sites​ /­ama​-­assn​.­org​/­files​/­corp​/­media​-­browser​/­public​/­arc​-­public​/­state​-­law​-­physician​-­assistant​-­scope​

360  •  Notes to Chapter 9 -­practice​.­pdf (accessed May 23, 2019); “In­de­pen­dent Practice of Medicine by Advanced Registered Nurses H-35.988,” https://­policysearch​.­ama​-­assn​.­org​/­policyfinder​/­detail​/­aprn%20compact​?­uri​=­%2FAMADoc%2FHOD​.­xml​-­0​-­2995​.­xml (accessed May 23, 2019). 27. Edgar Walters, “Medical Authority Bills Spark More Doc Fights,” Texas Tribune, April 21, 2015, https://­w ww​.­texastribune​.­org​/­2015​/­0 4​/­21​/­doc​-­fighting​-­bills​-­would​-­change​ -­scope​-­medical​-­prac​/­. 28. “Quality Eye Care Defense Fund,” Texas Ophthalmological Foundation, http://­w ww​ .­texaseyes​.­org ​/­qedf​-­lobbying​-­f und (accessed June 4, 2019). 29. The Center for Responsive Politics, Top Spender, https://­w ww​.­opensecrets​.­org ​/­lobby​ /­top​.­php​?­showYear​=­2016&indexType​=­s (accessed September 29, 2017). 30. Ronald Hamowy, “The Early Development of Medical Licensing Laws in the United States,” Journal of Libertarian Studies 3 (1979): 73–119. 31. Paul J. Feldstein, Health Associations and the Demand for Legislation: The Po­liti­cal Economy of Health (Cambridge, MA: Ballinger Publishing, 1977), 59. 32. Christy Ford Chapin, Ensuring Amer­i­ca’s Health: The Public Creation of the Corporate Health Care System (Cambridge: Cambridge University Press, 2015). 33. Christine C. Kushner, The Feasibility of Health Care Cooperatives in Rural Amer­i­ca: Learning from the Past to Prepare for the F ­ uture (Chapel Hill: University of North Carolina Health Ser­vices Research Center, 1991), https://­w ww​.­shepscenter​.­unc​.­edu​/­rural​/­pubs​/­report​ /­W P12​.­pdf. 34. Data on income are from U.S. Bureau of Economic Analy­sis, National Income and Product Accounts, “­Table 2.1. Personal Income and Its Disposition,” https://www.bea.gov/ (accessed June 19, 2019). I divided “personal income” by population to calculate average incomes. Data on health insurance premiums for 2018 are reported by the National Conference of State Legislatures, Health Insurance Premiums and Increases, December 4, 2018, http://­w ww​.­ncsl​.­org ​/­research​/­health​/­health​-­insurance​-­premiums​.­aspx. 35. Kushner, The Feasibility of Health Care Cooperatives in Rural Amer­i­ca. 36. Ford Chapin, Ensuring Amer­i­ca’s Health. 37. Paul Starr, The Social Transformation of American Medicine: The Rise of a Sovereign Profession and the Making of a Vast Industry (New York: Basic Books, 2008). 38. “Indictment of the American Medical Association for Activities in Washington,” New York Times, December 21, 1938. 39. J. Brosig-­Koch, H. Hennig-­Schmidt, N. Kairies-­Schwarz, and D. Wiesen, “The Effects of Introducing Mixed Payment Systems for Physicians: Experimental Evidence,” Health Economics 26.2 (2017): 243–62. 40. Ford Chapin, Ensuring Amer­i­ca’s Health. 41. Karen Smiley, Medical Billing & Coding for Dummies, 2nd ed. (Hoboken, NJ: Wiley and Sons, 2015). 42. Bureau of L ­ abor Statistics, Occupational Employment Statistics, Occupational Employment and Wages, May 2018, 29-2071: Medical Rec­ords and Health Information Technicians, https://­w ww​.­bls​.­gov​/­oes​/­current​/­oes292071​.­htm. 43. Bureau of ­Labor Statistics, Occupational Employment Statistics, Occupational Employment and Wages, May 2018, 43-3021: Billing and Posting Clerks, https://­w ww​.­bls​.­gov​ /­oes​/­current​/­oes433021​.­htm. 44. It should be said that t­ hese numbers understate the a­ ctual numbers of billing professionals and doctors, b­ ecause the Occupational Employment Statistics from which they are derived exclude ­owners of unincorporated businesses, which describes many doctors, and the self-­employed, who contract out their ser­vices in the case of rec­ords keepers (U.S. Bureau of

Notes to Chapter 9  •  361 ­ abor Statistics, Occupational Employment Statistics, https://­w ww​.­bls​.­gov​/­oes​/­home​.­htm L [accessed May 30, 2019]). Census data from 2016 show that t­ here are approximately one million physicians and surgeons. The ratio of administrators to physicians is prob­ably still around 1 to 2, but even 1 to 3 would be excessive. 45. For relevant discussion, see Jonathan Rothwell, “No Recovery: An Analy­sis of Long-­ term Productivity Decline” (Washington, DC: Gallup Organ­ization, 2017); C. Peterson and R. Burton, “U.S. Health Care Spending: Comparison with Other OECD Countries” (Washington, DC: Congressional Research Ser­vice, 2007). 46. D. Morra, S. Nicholson, W. Levinson, D. Gans, T. Hammonds, and L. Casalino, “U.S. Physician Practices versus Canadians: Spending Nearly Four Times as Much Money Interacting with Payers,” Health Affairs 30.8 (2011): 1443–50; J. Minnot, “What Are the Costs to Physicians of Administrative Complexity in Their Interactions with Payers?” Findings Brief: Health Care Financing & Organ­ization 13.2 (2010): 1–3. 47. A. Jiwani, D. Himmelstein, S. Woolhandler, and J. G. Kahn, “Billing and Insurance-­ Related Administrative Costs in United States’ Health Care: Synthesis of Micro-­Costing Evidence,” BMC Health Ser­ vices Research 14 (2014): 556, https://­bmchealthservres​ .­biomedcentral​.­com​/­articles​/­10​.­1186​/­s12913​-­014​-­0556​-­7; E. Wikler, P. Basch, and D. Cutler, Three Strategies for Reducing Health Care Administrative Costs (Washington, DC: Center for American Pro­g ress, 2012). 48. Centers for Medicare and Medicaid Ser­vices, 2016 National Health Expenditure Accounts, https://­w ww​.­cms​.­gov​/­Research​-­Statistics​-­Data​-­and​-­Systems​/­Statistics​-­Trends​-­and​ -­Reports​/ ­NationalHealthExpendData​/ ­NationalHealthAccountsHistorical​.­html. 49. M. H. Michal, M. S. Pekarske, M. K. Mc­Manus, and R. Van Deuren, “Corporate Practice of Medicine Doctrine: 50 State Survey Summary” (Madison, WI: Center to Advance Palliative Care and National Hospice and Palliative Care Organ­ization, 2006). 50. OECD, “Health Workforce Policies in OECD Countries: Right Jobs, Right Skills, Right Places” (Paris: OECD, 2016), http://­w ww​.­oecd​.­org ​/­publications​/­health​-­workforce​ -­policies​-­in​-­oecd​-­countries​-­9789264239517​-­en​.­htm. 51. Canadian Medical Protective Association, “Physicians and Nurse Prac­ti­tion­ers: Working Collaboratively as In­de­pen­dent Health Professionals,” 2014, https://­w ww​.­cmpa​-­acpm​ .­c a​/­en​/­a dvice​-­publications​/­browse​-­a rticles​/­2 014​/­physicians​-­a nd​-­nurse​-­practitioners​ -­working​-­collaboratively​-­as​-­independent​-­health​-­professionals. 52. Australian Medical Association, “In­de­pen­dent Nurse Prac­ti­tion­ers,” 2005, https://­ ama ​.­com​.­au​/­s ystem​/­tdf​/­documents​/­A MA ​_ ­I ndependent​_ ­Nurse ​_ ­Practitioner​_ ­Position​ _ ­Statement​.­pdf​?­fi le​=­1&type​=­node&id​=­4 0709; J. Carvel, “Nurses to Get Far-­Reaching Prescribing Powers,” Guardian, November  10, 2005, 4. For further discussion, see S. Elsom, B. Happell, and E. Manias, “Nurse Prac­ti­tion­ers and Medical Practice: Opposing Forces or Complementary Contributions?” Perspectives in Psychiatric Care 45.1 (2009): 9–16. 53. H. Eckstein, Pressure Group Politics: The Case of the British Medical Association (Stanford, CA: Stanford University Press, 1960), 92. 54. N. Ketel, E. Leuven, H. Oosterbeek, and B. Klaauw, “Do Dutch Dentists Extract Mono­poly Rents?,” Journal of Health Economics 63 (2019): 145–58. 55. Morris M. Kleiner and Kyoung Won Park, “­Battles among Licensed Occupations: Analyzing Government Regulations on ­Labor Market Outcomes for Dentists and Hygienists” (Cambridge, MA: NBER Working Paper 16560, 2010). 56. M. Langelier, B. Baker, and T. Continelli, Development of a New Dental Hygiene Professional Practice Index by State, 2016 (Rensselaer, NY: Oral Health Workforce Research

362  •  Notes to Chapter 9 Center, Center for Health Workforce Studies, School of Public Health, State University of New York, Albany, November 2016). 57. Oral Health Workforce Research Center, “Variation in Dental Hygiene Scope of Practice by State,” http://­w ww​.­oralhealthworkforce​.­org​/­resources​/­variation​-­in​-­dental​-­hygiene​ -­scope​-­of​-­practice​-­by​-­state/ (accessed May 30, 2019). 58. David A. Nash, Jay W. Freedman, Kavita R. Mathu-­Muju, Peter G. Robinson, Julie Satur, Susan Moffat, Rosemary Kardos, et al., “A Review of the Global Lit­er­a­ture on Dental Therapists,” Community Dentistry and Oral Epidemiology 4 2.1 (2014): 1–10. 59. Jay W. Friedman and Kavita R. Mathu-­Muju, “Dental Therapists: Improving Access to Oral Health Care for Underserved ­Children,” American Journal of Public Health 104.6 (2014): 1005–9; Mary Jordan, “The Unexpected Po­liti­cal Power of Dentists,” Washington Post, July 1, 2017, https://­w ww​.­washingtonpost​.­com​/­politics​/­the​-­unexpected​-­political​-­power​ -­of​-­dentists​/­2017​/­07​/­01​/­ee946d56​-­54f3​-­11e7​-­a 204​-­ad706461fa4f​_­story​.­html​?­noredirect​ =­on&utm​_­term​=­​.­57acfb330cc9. 60. United States Supreme Court, North Carolina State Board of Dental Examiners v. Federal Trade Commission, 2015, https://­w ww​.­ftc​.­gov​/­system​/­fi les​/­documents​/­cases​/­150225nc​ dentalopinion​.­pdf. 61. For the first set of evidence, see chapter 5. For the advanced analy­sis, see John Abowd, Francis Kramarz, Paul Lengermann, Kevin McKinney, and Sébastien Roux, “Per­sis­tent Inter-­ Industry Wage Differences: Rent Sharing and Opportunity Costs,” IZA Journal of L ­ abor Economics 1.1 (2012): 1–25. 62. Luxembourg Income Study (LIS) Database, 2019, http://www.lisdatacenter.org. 63. OECD, “Competition in Professional Ser­vices.” 64. Iain Paterson, Marcel Fink, and Anthony Ogus, Economic Impact of Regulation in the Field of Liberal Professions in Dif­fer­ent Member States (Brussels: CEPS, 2003); OECD, “Competitive Restrictions in L ­ egal Professions 2007,” Directorate for Financial and Enterprise Affairs Competition Committee, DAF/COMP (2007) 39 (2009), https://­w ww​ .­oecd​.­org ​/­regreform​/­sectors​/­4 0080343​.­pdf. “In Finland, non-­lawyers could represent clients in court but since 2002 a law degree is required to perform ­those ser­vices” (Stockholm Institute for Scandinavian Law, “Finnish Bar Association,” 52, http://­w ww​.­scandinavianlaw​ .­se​/­pdf​/­4 6​-­15​.­pdf. [accessed May 30, 2019]). 65. Paterson, Fink, and Ogus, Economic Impact of Regulation. 66. National Conference of Bar Examiners, 2015 Statistics, http://­w ww​.­ncbex ​.­org​ /­dmsdocument​/­195 (accessed May 31, 2019). 67. Society of American Law, “Teachers Statement on the Bar Exam, July 2002,” Journal of L ­ egal Education 52.3 (2002): 446–52. 68. Seth Williams, “Closing Real Estate Deals with an Attorney—­W hen & Where Is It Required?” REtipster, https://­retipster​.­com​/­real​-­estate​-­closing​-­a gents/ (accessed November 2, 2017). 69. Goldfarb v. ­Virginia State Bar, 421 U.S. 773, 788–789, n. 17, 95 S.Ct. 2004, 2013, 44 L.Ed.2d 572 (1975). 70. The American Bar Association, Model Rules of Professional Conduct, 5.4, https://­ www​.­a mericanbar​.­org ​/­g roups​/­professional ​_­responsibility​/­publications​/­model ​_­r ules ​_­of​ _­professional​_­conduct​/­rule​_­5​_­4 ​_ ­professional​_­independence​_­of​_­a ​_ ­lawyer​.­html (accessed May 31, 2019). 71. ABA Commission on the F ­ uture of L ­ egal Ser­vices, “For Comment: Issues Paper Regarding Alternative Business Structures,” April  8, 2016, http://­abafuturesreport​.­com​ /­#download​-­f ull​-­report.

Notes to Chapter 9  •  363 72. ABA Commission on the ­Future of ­Legal Ser­vices, “For Comment.” 73. Gallup, Honesty/Ethics in Professions, December 3–12, 2018, and December 4–11, 2017, http://­news​.­gallup​.­com​/­poll​/­1654​/­Honesty​-­Ethics​-­Professions​.­a spx. (Item states, “Please tell me how you would rate the honesty and ethical standards of ­people in ­these dif­ fer­ent fields—­very high, high, average, low, or very low?”) 74. American Bar Association Consortium on ­Legal Ser­vices and the Public, “­Legal Needs and Civil Justice: A Survey of Americans,” 1994, https://­w ww​.­americanbar​.­org​/­content​/­dam​ /­a ba​/­a dministrative​/­l egal​_­a id​_­i ndigent​_­d efendants​/­d ownloads​/­l egalneedstudy​ .­authcheckdam​.­pdf. 75. Washington State Supreme Court Task Force on Civil Equal Justice Funding, “The Washington State Civil ­L egal Needs Study,” 2003, http://­w ww​.­courts​.­wa​.­gov​/­newsinfo​ /­content​/­taskforce​/­civillegalneeds​.­pdf. 76. Washington State Supreme Court Task Force on Civil Equal Justice Funding, “2015 Washington State Civil L ­ egal Needs Study Update,” http://­ocla​.­wa​.­gov​/­wp​-­content​/­uploads​ /­2015​/­10​/­CivilLegalNeedsStudy ​_­October2015​_­V21​_­Final10​_­14 ​_­15​.­pdf. 77. Solicitors Regulation Authority, “Improving Access—­Tackling Unmet L ­ egal Needs,” June 2017, https://­w ww​.­sra​.­org​.­u k​/­risk​/­resources​/­legal​-­needs​.­page#. 78. Lorelei Laird, “Starved of Money for Too Long, Public Defender Offices Are Suing—­ and Starting to Win,” ABA Journal, January 2017, http://­w ww​.­abajournal​.­com​/­magazine​ /­article​/­the​_ ­g ideon​_­revolution; “Attorney General Eric Holder Speaks at the American Bar Association’s National Summit on Indigent Defense” (Washington, DC: U.S. Department of Justice, February  4, 2012), https://­w ww​.­justice​.­gov​/­opa​/­speech​/­attorney​-­general​-­eric​ -­holder​-­speaks​-­american​-­bar​-­association​-­s​-­national​-­summit​-­indigent. For facts on overburden of case loads, see Donald J. Farole and Lynn Langton, County-­Based and Local Public Defender Offices, 2007 (Washington, DC: U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics, 2010). 79. ­Limited Practice Officers, https://­w ww​.­wsba​.­org​/­for​-­legal​-­professionals​/­join​-­the​-­legal​ -­profession​-­in​-­wa​/­limited​-­practice​-­officers (accessed May  31, 2019); Washington Supreme Court, Admission and Practice Rule 12, “­Limited Practice Rule for ­Limited Practice Officers,” https://­w ww​.­courts​.­wa ​.­gov​/­court​_­r ules​/­​?­fa​=­court​_­r ules​.­d isplay&group​=­g a&set​ =­apr&ruleid​=­gaapr12 (accessed May 31, 2019); Williams, “Closing Real Estate Deals with an Attorney.” 80. Scott A. Smith, Juan Pablo Paredes, Rebecca M. Baker, and Mark Hanku Kim, “Practice of Law Board Resignations,” a letter to the Washington State Supreme Court, November  9, 2015, http://­w ww​.­abajournal​.­com​/­fi les​/­L etter​_­to​_ ­Supreme ​_­Court​_ ­E xplaining​ _­Resignations​.­pdf. 81. Robert Ambrogi, “Washington State Moves around UPL, Using L ­ egal Technicians to Help Close the Justice Gap,” ABA Journal, January 2015, http://­w ww​.­abajournal​.­com​ /­magazine​/­a rticle​/­washington ​_ ­state ​_ ­moves ​_ ­a round ​_­upl ​_­u sing ​_ ­legal ​_­technicians ​_­to​ _­help​_­close​_­the. 82. Ambrogi, “Washington State Moves around UPL.” 83. Washington State Court Rules, Admission and Practice Rules, Rule 28, L ­ imited Practice Rule for L ­ imited License ­Legal Technicians, http://­w ww​.­courts​.­wa ​.­gov​/­court​_­rules​/­​ ?­fa​=­court​_­rules​.­list&group​=­ga&set​=­apr (accessed May 31, 2019). 84. U.S. Bureau of L ­ abor Statistics, Occupational Employment Statistics, 2017, https://­ www​.­bls​.­gov​/­oes​/­current​/­oes​_­42660​.­htm#23​-­0 000. 85. Smith et  al., “Practice of Law Board Resignations”; Victor Li, “Board Members Quit, Blast Washington State Bar in Fight over UPL, L ­ egal Technicians,” ABA Journal,

364  •  Notes to Chapter 9 November 9, 2015, http://­w ww​.­abajournal​.­com​/­news​/­article​/­Board ​_­members ​_ ­quit​_ ­blast​ _­Washington​_ ­State ​_ ­Bar​_­in​_ ­fight​_­over​_­U PL. 86. Smith et al., “Practice of Law Board Resignations.” 87. Thomas M. Clarke and Rebecca L. Sandefur, “Preliminary Evaluation of the Washington State ­Limited License ­L egal Technician Program,” March 2017, p. 9, http://­w ww​ .­a mericanbarfoundation​.­org ​/­uploads​/­cms​/­documents​/­preliminary ​_­e valuation ​_­of​_­t he​ _­washington​_­state​_­limited​_­license​_­legal​_­technician​_ ­program​_­032117​.­pdf. 88. Washington State Bar Association, ­L egal Directory, https://­w ww​.­mywsba​.­org​ /­LegalDirectory​.­aspx (accessed April 16, 2018). 89. My analy­sis of U.S. Bureau of Economic Analy­sis, ­table 6.2D, “Compensation of Employees by Industry” and more recent versions of the same t­ able, as well as t­ able 6.5B, “Full-­Time Equivalent Employees by Industry,” and its more recent time periods, https://www.bea.gov/. 90. My analy­sis of IPUMS USA 2012–2016 American Community Survey, using all U.S. residents 18 and older and indnaics. Tax data analyzed by Jon Bakija, Adam Cole, and Bradley T. Heim, “Jobs and Income Growth of Top Earners and the ­Causes of Changing Income In­equality: Evidence from US Tax Return Data,” unpublished manuscript, Williams College, 2012, who also find that ­those in medical occupations, ­lawyers, and financial professionals constitute a larger fraction of the top 1 ­percent of income earners. They find that 15.7 ­percent of top income earners are medical workers, 13.9 ­percent are financial professionals and man­ag­ers, and 8.4 ­percent are ­lawyers. 91. This divides the total number of employees (34,400) by total 2016 compensation and benefits: $11.6 billion. Goldman Sachs, 2016 Annual Report, http://­w ww​.­goldmansachs​.­com​ /­investor​-­relations​/­fi nancials​/­current​/­a nnual​-­reports​/­2016​-­a nnual​-­report​/­a nnual​-­report​ -­2016​.­pdf. 92. Jacob Sonenshine, “Hedge Fund Man­ag­ers Are Celebrating Big Pay Increases—­Here’s How Much You Can Expect to Earn Working at One,” Business Insider, January 31, 2018, http://­w ww​.­businessinsider​.­com​/­hedge​-­f und​-­pay​-­is​-­going​-­up​-­2018​-­1. 93. Jonathan Marino, “How Goldman Sachs Makes Money on Its Top Traders—­A fter They Quit,” Business Insider, May 12, 2015. 94. Usha Rodrigues, “Securities Law’s Dirty L ­ ittle Secret,” Fordham Law Review 81 (2012): 3389–3437. 95. Abowd et al., “Per­sis­tent Inter-­Industry Wage Differences”; Brink Lindsey and Steven Teles, The Captured Economy: How the Power­ful Enrich Themselves, Slow Down Growth, and Increase In­equality (Oxford: Oxford University Press, 2017). 96. Randall S. Kroszner and Philip E. Strahan, “What Drives Deregulation? Economics and Politics of the Relaxation of Bank Branching Restrictions,” Quarterly Journal of Economics 114.4 (1999): 1437–67. 97. Jith Jayaratne and Philip E. Strahan, “The Finance-­Growth Nexus: Evidence from Bank Branch Deregulation,” Quarterly Journal of Economics 111.3 (1996): 639–70; P. E. Strahan and D. C. Wheelock, “The Real Effects of US Banking Deregulation,” commentary, Review—­Federal Reserve Bank of St. Louis 85.4 (2003): 111–28. 98. Vanguard Mutual Funds, https://­investor​.­vanguard​.­com​/­mutual​-­f unds​/­list#​/­mutual​ -­f unds​/­asset​-­class​/­month​-­end​-­returns (accessed June 4, 2019). 99. Alan Rappeport, “A Short History of Hedge Funds,” CFO, March  27, 2007, http://­w w2​.­cfo​.­com​/­banking​-­capital​-­markets​/­2007​/­03​/­a​-­short​-­history​-­of​-­hedge​-­f unds​/­. 100​.­U.S. Securities and Exchange Commission, “Updated Investor Bulletin: Accredited Investors,” Investor.Gov, January 31, 2019, https://www.investor.gov/additional-re​ sources/news-alerts/alerts-bulletins/updated-investor-bulletin-accredited-investors.

Notes to Chapter 10  •  365 101. Cary Martin Shelby, “Privileged Access to Financial Innovation,” Loyola University Chicago Law Journal 47 (2015): 315. 102. Alexander J. Davie, “Accredited Investors vs. Qualified Clients vs. Qualified Purchasers: Understanding Investor Qualifications,” Strictly Business, blog, August  17, 2017, https://­w ww​.­strictlybusinesslawblog​.­com​/­2 017​/­0 8​/­17​/­accredited​-­i nvestors​-­vs​-­qualified​ -­clients​-­vs​-­qualified​-­purchasers​/­. 103. Shelby, “Privileged Access to Financial Innovation.” 104. KPMG International, “The Value of the Hedge Fund Industry to Investors, Markets, and the Broader Economy” (The Center for Hedge Fund Research, Imperial College London, 2012). 105. “What Are the Hedge Fund Marketing Rules?,” Deke Digital, https://­w ww​.­dekedigital​ .­com​/­marketing ​/­what​-­are​-­the​-­hedge​-­f und​-­marketing​-­rules​/­(accessed June 4, 2019). 106. U.S. Securities and Exchange Commission, Division of Investment Management, “Private Funds Statistics First Calendar Quarter 2017,” ­table  15, https://­w ww​.­sec​.­gov​ /­divisions​/­investment​/­private​-­f unds​-­statistics​/­private​-­f unds​-­statistics​-­2017​-­q1​.­pdf. 107. Nassim Nicholas Taleb, Skin in the Game: Hidden Asymmetries in Daily Life (New York: Random House, 2018). 108. U.S. Securities and Exchange Commission, Division of Investment Management, “Private Funds Statistics First Calendar Quarter 2017,” ­table 14. 10. CREATING A JUST SOCIET Y 1. Eric Alden Smith, Kim Hill, Frank W. Marlowe, David Nolin, Polly Wiessner, Michael Gurven, Samuel Bowles, et  al., “Wealth Transmission and In­equality among Hunter-­ Gatherers,” Current Anthropology 51.1 (2010): 19–34. 2. Robert J. Gordon, The Rise and Fall of American Growth: The US Standard of Living since the Civil War (Prince­ton, NJ: Prince­ton University Press, 2017). 3. Mancur Olson, The Rise and Decline of Nations: Economic Growth, Stag flation, and Social Rigidities (New Haven, CT: Yale University Press, 2008). 4. Brink Lindsey and Steven Teles, The Captured Economy: How the Power­ful Enrich Themselves, Slow Down Growth, and Increase In­equality (New York: Oxford University Press, 2017). 5. Executive Order 13563—­I mproving Regulation and Regulatory Review, January 8, 2011, https://­w ww​.­whitehouse ​.­gov​/­t he​-­press​-­office​/­2 011​/­01​/­18​/­executive​-­order​-­13563​ -­improving​-­regulation​-­and​-­regulatory​-­review. 6. Michael Mandel and Diana G. Carew, “Regulatory Improvement Commission: A Politically-­Viable Approach to U.S. Regulatory Reform” (Washington, DC: Progressive Policy Institute, 2013); S.708 Regulatory Improvement Act, https://­w ww​.­congress​.­gov​/­bill​ /­114th​-­congress​/­senate​-­bill​/­708​/­text​?­format​=­txt (accessed May 31, 2019). 7. Kevin R. Kosar, “Is Regulatory Reform a Hopeless Cause?” (Washington, DC: The Brookings Institution, 2015). 8. Ida Tarbell, The History of the Standard Oil Com­pany (New York: McClure, Phillips, and Com­pany, 1904), http://­w ww​.­pagetutor​.­com​/­standard​/­chapter16​_ ­part6​.­html. 9. Federal Trade Commission, Annual Report, “Enforcement,” https://­w ww​.­ftc​.­gov​ /­reports​/­annual​-­highlights​-­2015​/­enforcement (accessed May 31, 2019). 10. Beth Redbird, “The New Closed Shop? The Economic and Structural Effects of Occupational Licensure,” American So­cio­log­i­cal Review 82.3 (2017): 600–624.

366  •  Notes to Chapter 10 11. Jonathan Rothwell, “No Recovery: An Analy­sis of Long-­Term U.S. Productivity Decline” (Washington, DC: Gallup, 2016), https://­news​.­gallup​.­com​/­reports​/­198776​/­no​ -­recovery​-­analysis​-­long​-­term​-­productivity​-­decline​.­aspx. 12. Rana Foroohar, Makers and Takers: The Rise of Finance and the Fall of American Business (New York: Crown Books, 2016). 13. Anat Admati and Martin Hellwig, The Bankers’ New Clothes: What’s Wrong with Banking and What to Do about It (Prince­ton, NJ: Prince­ton University Press, 2014). 14. Houman B. Shadab, “Fending for Themselves: Creating a US Hedge Fund Market for Retail Investors,” NYU Journal of Legislature and Public Policy 11 (2007): 251–319. 15. Mark A. Lemley, “Software Patents and the Return of Functional Claiming,” Stanford Public Law Working Paper No. 2117302, October 12, 2012, https://­ssrn​.­com​/­abstract​ =­2117302 or http://­d x​.­doi​.­org​/­10​. ­2139​/­ssrn​. ­2117302. 16. James Bessen and Michael James Meurer, Patent Failure: How Judges, Bureaucrats, and L ­ awyers Put Innovators at Risk (Prince­ton, NJ: Prince­ton University Press, 2008). 17. Mustaqeem Siddiqui and S. Vincent Rajkumar, “The High Cost of Cancer Drugs and What We Can Do about It,” Mayo Clinic Proceedings 87.10 (2012): 935–43; Vincent Rajkumar, “On the High Price of Cancer Drugs,” Library of Economics and Liberty, April 9, 2018, http://­w ww​.­econtalk​.­org​/­archives​/­2018​/­0 4​/­vincent​_­rajkuma​.­html. 18. Jonathan T. Rothwell, “The Effects of Racial Segregation on Trust and Volunteering in US Cities,” Urban Studies 49.10 (2012): 2109–36; Thomas F. Pettigrew and Linda R. Tropp, “A Meta-­A nalytic Test of Intergroup Contact Theory,” Journal of Personality and Social Psy­chol­ogy 90.5 (2006): 751–83. 19. Timothy J. Bartik, “From Preschool to Prosperity: The Economic Payoff to Early Childhood Education” (Kalamazoo, MI: W.E. Upjohn Institute for Employment Research, 2014), https://­doi​.­org​/­10​.­17848​/­9780880994835. 20. Jonathan Rothwell, “No Recovery: An Analy­sis of Long-­Term Productivity Decline” (Washington, DC: Gallup 2016), https://news.gallup.com/reports/198776/no-recovery​ -analysis-long-term-productivity-decline.aspx. 21. OECD, Table D3.2a, “Teachers’ Actual Salaries Relative to Earnings of TertiaryEducated Workers (2015),” in Education at a Glance 2017: OECD Indicators (Paris: OECD, 2017), http://­d x​.­doi​.­org​/­10​.­1787​/­eag​-­2017​-­en; OECD, “Indicator D3: How Much Are Teachers Paid?” Education at a Glance 2014: OECD Indicators (Paris: OECD, 2014), http://­d x​.­doi​.­org​/­10​.­1787​/­888933119815 (accessed May  31, 2019); chart D3.1, “Teachers’ Salaries Relative to Earnings for Tertiary-­Educated Workers Aged 25–64 (2012).” 22. Rothwell, “No Recovery.” 23. Jonathan Rothwell, The Hidden STEM Economy (Washington, DC: Metropolitan Policy Program at Brookings, 2013), https://www.brookings.edu/research/the-hidden​-stem​-economy/. 24. Phillip Toner, “Tradespeople and Technicians in Innovation,” in Fostering Enterprise: The Innovation and Skills Nexus—­Research Readings, ed. Penelope Curtin, John Stanwick, and Francesca Beddie (Adelaide: Australian Government, National Center for Vocational Education Research, 2012); Phillip Toner, Tim Turpin, and Richard Woolley, “The Role and Contribution of Tradespeople and Technicians in Australian Research and Development: An Exploratory Study” (Centre for Industry and Innovation Studies, University of Western Sydney, 2011). 25. V ­ irginia Beach City Public Schools, “21st  ­Century Industry Credentials: An ­A nnual Report of School Per­for­mance,” http://­w ww​.­vbschools​.­com​/­TCE​/­content​/­pdfs​ /­WorkplaceReadinessSkills​.­pdf (accessed May 10, 2018).

Notes to Appendix  •  367 26. Mathilde Almlund, Angela Lee Duckworth, James Heckman, and Tim Kautz, “Personality Psy­chol­ogy and Economics,” in Handbook of the Economics of Education (Elsevier, 2011), 4:1–81; Falk Leichsenring and Sven Rabung, “Effectiveness of Long-­Term Psychodynamic Psychotherapy: A Meta-­A nalysis,” Journal of the American Medical Association 300.13 (2008): 1551–65. 27. Martin R. West, “Should Non-­Cognitive Skills Be Included in School Accountability Systems? Preliminary Evidence from California’s CORE Districts,” Evidence Speaks Reports 1.13 (2016), https://­w ww​.­brookings​.­edu​/­research​/­should​-­non​-­cognitive​-­skills​ -­be​-­included​-­in​-­school​-­accountability​-­systems​-­preliminary​-­evidence​-­from​-­californias​-­core​ -­districts​/­. APPENDIX 1. World In­equality Database, https://­wid​.­world​/­(accessed June 5, 2019). 2. IPUMS USA, https://­usa​.­ipums​.­org​/­usa​/­(accessed June 5, 2019). 3. Luxembourg Income Study, https://­w ww​.­lisdatacenter​.­org​/­(accessed June 5, 2019). 4. Thomas Piketty, Emmanuel Saez, and Gabriel Zucman, Distributional National Accounts: Methods and Estimates for the United States, no. w22945 (Cambridge, MA: National Bureau of Economic Research, 2016). 5. American Acad­emy of F ­ amily Physicians, https://­w ww​.­udsmapper​.­org​/­zcta​-­crosswalk​ .­cfm (accessed May 25, 2019). 6. Missouri Census Data Center, Geocorr 2014: Geographic Correspondence Engine, http://­mcdc​.­missouri​.­edu​/­websas​/­geocorr14​.­html (accessed May 25, 2019). 7. Robert C. Feenstra, Robert Inklaar, and Marcel P. Timmer, “The Next Generation of the Penn World ­Table,” American Economic Review 105.10 (2015): 3150–82, www​.­ggdc​.­net​/­pwt. 8. Robert Carroll, “Income Mobility and the Per­sis­tence of Millionaires, 1999 to 2007,” Tax Foundation Special Report No. 180 (Washington, DC, 2010), https://­taxfoundation​.­org​ /­article​/­income​-­mobility​-­and​-­persistence​-­millionaires​-­1999​-­2007. 9. Life evaluation is mea­sured by the following item on the Gallup World Poll: “Please imagine a ladder, with steps numbered from 0 at the bottom to 10 at the top. The top of the ladder represents the best pos­si­ble life for you and the bottom of the ladder represents the worst pos­si­ble life for you. On which step of the ladder would you say you personally feel you stand at this time?” (https://­w ww​.­gallup​.­com​/­a nalytics​/­232838​/­world​-­poll​.­a spx [accessed June 5, 2019]). 10. The World Bank, “GDP per Capita, PPP (Current International $),” https://­data​ .­worldbank​.­org​/­indicator​/ ­N Y​.­GDP​.­PCAP​.­PP​.­CD (accessed June 5, 2019). 11. Charles I. Jones, “The Facts of Economic Growth,”in Handbook of Macroeconomics (Elsevier, 2016), 2: 3–69; Chang-­Tai Hsieh, Erik Hurst, Charles I. Jones, and Peter J. Klenow, The Allocation of Talent and US Economic Growth, no. w18693 (Cambridge, MA: National Bureau of Economic Research, 2013). 12. For a related argument, see Kevin M. Murphy, Andrei Shleifer, and Robert W. Vishny, “The Allocation of Talent: Implications for Growth,” Quarterly Journal of Economics 106.2 (1991): 503–30.

INDEX

Page numbers in italics refer to figures and ­tables. “Accept No Substitute,” 247 Admati, Anat, 280 adoption, 184–­85, 341n68; noncognitive skills and, 88, 90; twin studies of, 141 African Americans and Hispanics: access to markets, 101; educational attainment of, 144; housing barriers for, 191, 201; income in­equality of, 45; income mobility of, 221; Industrial Revolution role, 163–67; intellectual achievements of, 291; IQ scores of, 84, 152–53, 173, 178, 283, 284; po­liti­cal in­equality and, 274; school readiness of, 186–87. See also housing segregation; Jim Crow Alesina, Alberto, 46–47 Alli, Dee, 190 Alternative for Germany Party (AfD), 3 altruism, 53, 109; ge­ne­tic basis, 57–58, 69 American Bar Association (ABA), 259–64; African Americans barred from, 101; bar exam deficiencies, 261–62; Model Rules of Professional Conduct, 262; monopolistic practices of, 262; quality information lacking for, 261 American Bar Foundation, 265 American Community Survey (ACS): home values, 301, 324n95; occupational data, 331n20; top earnings, 20, 296–99 American Dental Association, 257–58 American Hospital Association, 251

American Medical Association (AMA), 248–53, 255–57; African Americans barred from, 101; antitrust lawsuit, 253; medical school accreditation and, 251; state licensing role of, 251 antipoverty policies, 106, 109, 226, 288–89 Aristotle, 52, 148–49 Armed-­Forces Qualification Test (AFQT), 72–73 Armed Ser­vices Vocational Aptitude Battery (ASVAB), 81, 185, 330n10; black-­white disparity of, 331n19 Asian American success theories: ­family size theory, 180; Harvard discrimination suit, 343n18; IQ scores and, 347n89; parents’ education and, 184–85; professions and occupations and, 160–61, 166; test scores of, 152, 154, 175, 177 Atkinson, Robert, 103, 106 Australian Medical Association, 256 Auten, Gerald, 317n9 Autor, David, 18 Avery, Derek, 94 Baker, Henry, 164 Baker, Thomas, 196 Baldwin, James, 51 Bannon, Steve, 48

370 • Index Bartles, Larry, 36 basic liberties, 62–63, 99–103 Beauchamp, Jonathan P., 142 Belgium: economic class segregation in, 216; physician compensation in, 244, 358n5; tax advice in, 259 Bell Curve, The (Murray, Herrnstein), 81, 118, 152–53, 175, 180 Benjamin, Miriam, 164 Berliner, Eli, 163 Bern­stein, Jared, 14 Berry, Charles, 171 “Big 5” personality traits, 73, 332n34 Bivens, Josh, 15 Blue Cross/Blue Shield, 250 Bogle, John, 269 Boone, Sarah, 164 Borjas, George, 11, 106–7 Botticini, Maristella, 156–58 Bowles, Samuel, 57, 181 Boyd, Marie, 208 Branigan, Amelia, 140 Bratsberg, Bernt, 127 Breitbart, 48 Brexit, 7 Brigham, Carl, 168–69, 171 British Medical Association, 256 Brodis, Christian, 10 Brosnan, Sarah, 54 Bureau of Economic Analy­sis, 267, 322n60 Bureau of ­Labor Statistics (BLS), 10, 72, 75, 85, 254 California Medical Association, 249–50 CalPERs, 14 capital income, 14–16, 31, 318n10; corporate profits and, 16, 321n44, 322n60; data analy­sis, 301, 303 Carver, George Washington, 164 Casciano, Rebecca, 229 Case, Anne, 220 Catholicism, preference option for the poor, 53, 64 Census Bureau, 12, 196, 221, 266, 308, 318n10 Center for Responsive Politics, 251

Chapin, Christy Ford, 252, 254 Chetty, Raj, 127, 222, 226 Chicago Child-­Parent Center, 132 China and Chinese companies, 103–4; educational attainment and, 177–78; entrepreneurship and, 104; immigrants from, 175; ­labor in, 9–10, 41, 105; trade and imports of, 12 Cicero, 53–54, 59, 64, 109, 326n10 Civil Rights Act (1964), 117 Civil Rights movement, 198, 274 CLASS, 128–29, 338nn35, 388n36 Clemens, Michael A., 11 Clinton, Hillary, 39, 121 Coates, Ta-­Nehisi, 192 Cochran, Gregory, 155–56, 159, 180 cognitive ability, 73–75, 77–80, 82, 84, 91–92. See also IQ cognitive ability testing, Programme for the International Assessment of Adult Competencies (PIAAC), 76–77, 322n71 cognitive ability test scores, Programme for the International Assessment of Adult Competencies (PIAAC), 177–79 cognitive and personality traits mea­sures. See Armed Ser­v ices Vocational Aptitude Battery (ASVAB); National Longitudinal Survey of Youth (NLSY); Programme for the International Assessment of Adult Competencies (PIAAC); Stanford-­Binet test Coleman, James, 117–20, 336n15 Coleman Report, 117–22, 175, 347n89 Coming Apart (Murray), 119 Community Property Own­er’s Protective Association, 198 compensation: low pay and, 71–72, 80; market-­based, 68–69; merit pay hy­po­thet­i­cal, 75–77; performance-­ based pay, 70; subjective evaluation, 70–71. See also predictors of income Congress Height Citizens Association, 199 Congressional Regulation Office, 276 conscientiousness, 91–92, 159, 292, 315, 332n34 Consumer Product Safety Commission, 228 Cook, Lisa D., 165

Index • 371 corporations, 91; domestic and foreign operations of, 15; power of, 13–17; salaries of, 25; taxation of, 13–15 Cruz, Ted, 39 culture: definition of, 181; effect of, 181; exposure to, 124; ­family and, 187, 220; intelligence and, 284–85; learning and, 184; neighborhood exposure and, 220 Current Population Survey, Voter Supplement, 36 Davis, Jefferson, 167 Demo­crats, racial views, 48, 50 Denton, Nancy, 202 Department of Health and ­Human Ser­vices, 237 Department of Housing and Urban Development, 202, 225 Detroit Public School District, 171–72 de Waal, Frans, 54 Diodorus Siculus, 157 Douglass, Frederick, 116–17, 167 drugs, sale or use, 237–41, 357nn48, 51 DuBois, W. E. B., 182–83 Durand, Jorge, 111 Dutch Health Care Authority, 257 Eckstein, Harry, 256 Eckstein, Zvi, 156–58 economic growth: economic slowdown and, 3–5, 7; income in­equality and, 5–7, 9, 275–76, 317n9 Economic Policy Institute, 15 education, in­equality in, 116; administrative costs, 286; ­career and technical education, 287; cognitive quality, 118, 145–46, 342n82, 342n84; effect on IQ, 123, 127, 132–33; environmental effects, 143–44; ­family and outside influences, 117–18, 121, 132, 286; income in­equality, 342n86; literacy for slaves, 117; race and IQ, 120, 128–32, 144–45, 336n15; Reconstruction era, 183–84; school and teacher quality, 120, 285–87; Southern re­sis­tance to educating blacks, 182. See also Coleman

Report; Flynn effect; ge­ne­tics; heredity and hereditarians; Proj­ect Talent; teachers; twin studies Eglin, Ellen, 165 Einstein, Katherine L., 208 Einthoven, Willem, 163 employee engagement, 92–95, 333nn41, 42, 43 employment: mass migration effect in, 11; trade, effect on, 10 entrepreneurship, 96, 137, 274; African Americans during Reconstruction, 167, 285; immigrants and, 11, 108; Marxism and public owner­ship, 64–65; patent trolls and, 281 Environmental Protection Agency, 228 Eu­ro­pean Social Survey Round 7, 41, 45–­47; immigration policy question, 324n96 Eurostat data, 358n5 Fair Housing Act of 1968, 202 fairness: evolutionary reasons for, 57–59; ­human experiments and, 55–57; innate sense of, 273; market-­based in­equality and, 113; nonhuman experiments and, 54; Rawls’ princi­ples and, 66–67; reciprocity and, 99, 273 Falconer’s formula, 139 Federal Housing Administration (FHA), 200 Federal Trade Commission, 258, 276 Fehr, Ernst, 55–56 Feldman, Stanley, 49 Fifth Ave­nue Association, 206 financial anxiety or security, 7, 319n16 Finland: cognitive per­for­mance in, 126; government, confidence in, 2; income in, 22, 24–25, 28; ­legal ser­vices in, 259, 362n64 Fischel, William, 204, 208 Flanagan, J. C., 172 Flynn effect, 126–27, 137; in K ­ enya, 123, 337n23; negative Flynn effects, 125–26 Foroohar, Rana, 279–80 Forscher, Patrick S., 48–49 Fox, Adam, 181

372 • Index ­free market: abuses of, 103–4; definition of, 102; fair competition, 104–5; innovation and, 105–6; international trade and, 104–5; preference to citizens and, 109–10; tariffs and, 103, 105; trade and, 112, 278. See also immigrants and immigration Freund, David M.P., 206 Friedman, John, 127 Fryer, Roland, 132, 236 Gächter, Simon, 55–56 Gallup Surveys: Builder Profile, 96; government, confidence in, 35; gradu­ate degrees, 154; Healthways Wellbeing Index, 231, 333n41; high or low income neighborhoods, 193; income class concerns, 37–39; ­lawyers’ honesty and ethics, 263; Minority Rights and Relations poll, 325nn102, 112; Q12,employee engagement, 92–93, 95, 333nn42, 43; racial views, 45; safety and security, 229; view of Trump, 41–42 Gallup World Poll: Asian test scores, 177; financial well-­being questions, 304, 367n9; government, confidence in, 2, 7, 8; housing satisfaction, 210, 353n55; questions asked in, 318n16 gender: business owner­ship and, 96; employee engagement and, 92–94; medical profession barriers and, 247; per­for­mance evaluations and, 92; ­women’s aptitude and, 86–87. See also Proj­ect Talent General Motors pension fund, 14 ge­ne­tics: environmental effects and, 142–43, 146, 283–84; genomic data and, 141–42; nucleotide polymorphisms (SNPs), 140, 142, 144–45, 342n85; polygenic score, 142–46, 342nn82, 85 Genome-­Wide Association Studies, 144 Germany: far right party of, 3; government, confidence in, 2, 7; home owner­ship in, 15; immigration and IQ in, 136; IQ of immigrants in, 136; Jews in, 160–61; physician top earners in, 28; zoning laws in, 216, 218

Gini coefficient, 303, 313, 322n61; pay in­equality, 77, 81, 331n16, 336n30 Gintis, Herbert, 57, 181 Glaeser, Ed, 204 Glick, David M., 208 globalization: expanded markets and, 19; income in­equality and, 272–73, 296, 308; ­labor income and, 13–14; superstar companies and, 21–22, 26, 29 golden rule, 52–53, 64; nonhuman experiments and, 54–55 Goldfarb, Michael, 158 Goldman Sachs, 266–67 Gordon, Robert, 34, 274 government, confidence in, 2–­3, 7–8, 36; data analy­sis, 295, 305–8; slow productivity and income in­equality, 35–36 government, lobbying and influence: campaign finance and, 249, 359n23; physicians’organ­izations and, 247–50, 256–57 government regulation, 276 Hamowy, Ronald, 251 Hardy, Jason, 155–56 Harpending, Henry, 155–56 Harris, Geneva, 233–34 Harris, Isaiah, 233 Harshbarger, David, 191 Harvard Opportunity Insights proj­ect, 231 Hayek, Frederick, 66 healthcare: benefits to society and, 34; dental hygienists limitations and, 257–58; dentists, top earners, 257; efficiency of, 32–34, 279; electronic medical rec­ords and, 322n70; entry barriers in, 245; inflation in, 37; nurse midwives in, 246–47; nurse prac­ti­tion­ ers in, 245–46, 249–50, 359n25; optometrists and, 250; top earners in, 24, 29, 31, 243–45, 255. See also physicians and surgeons healthcare, global: allied professions and, 256; dentist, quotas of, 257; educational requirements of, 256; national income share in, 30

Index • 373 healthcare financing: fee-­for-­service model, 253–55; group medical care model, 252–54, 278–79; insurance and insurers, 250, 251–52; insurance scrutiny in, 254–55; medical coding of, 254, 360–61n44 Hellwig, Martin, 280 Hendren, Nathaniel, 222 heredity and hereditarians, 85; biological dysgenic effects, 126, 180, 337n31; damage from, 291–93; debunked ideas of, 151–53; eugenics movement and, 150; ge­ne­tic differences, 85, 118–19, 155, 188–89; ge­ne­tic nurture, 144; Jewish success theories and, 156–63, 166–73, 180; northern Eu­ro­pean ancestry and, 150–51. See also ge­ne­tics; twin studies Herrnstein, Richard, 81, 118, 152–53, 180 Heywood, Alice, 181 housing: high or low income neighborhoods and, 193–­95; home values and, 302 housing, worldwide, 210–11 housing segregation: Asians and, 192; black-­white, 192, 201, 350n6; Chicago segregation, 198; exclusionary zoning and, 200, 203–4, 208; government action and, 191–92, 194–95, 282; Hispanics and, 192; income mea­sures and, 209; mortgage terms and, 202–3; red-­lining in, 200, 202; rental amounts, 196–97; restrictive covenants in, 198–200; vio­lence and intimidation in, 196–97; Washington D.C. segregation, 199. See also Federal Housing Administration (FHA); Jim Crow; lead exposure; zoning laws housing segregation, worldwide: French immigrants, 213–14; immigrants and immigration, 212–16, 353n63; inequity, U.S. and UK, 211–16; neighborhood satisfaction, 210–11; U.S. and Eu­rope, 216–18; U.S. and France, 214 Hoxby, Caroline, 119–20 Huddy, Leonie, 49, 306

Hult, Kristopher, 34 ­human capital, 18, 97, 163 Hunt, Jennifer, 11 hunter-­gather socie­ties, 58, 113, 273, 328n31 Hurst, Erik, 306 Hyde Park Improvement Association, 198 Imafidon, Ann Marie, 190 Imafidon, Chris, 190 immigrants and immigration: benefits and downsides of, 11–12, 106, 112; border security and, 110–11; documented and undocumented, 110–11; educational attainment of, 176, 334–35n18; effect of quality educational systems, 133–38; effect on income in­equality on, 110, 334n19; En­g lish national exams, 134–35; high or lesser skilled, 107–8, 111, 334–35n18; limits on, 108–9 income distribution, 39, 77, 82, 91, 108, 274; equality of, 15; savings, borrowing and, 10; slowdown of, 34; top one ­percent and, 19 income distribution, data, 210, 223 income distribution, global, 28, 259 income distribution, national, 41, 46 income growth, 318n10, 318n12, 318n15 income in­equality: access to hedge, private equity and venture capital funds, 280; data analy­sis, 296–305, 310–15; effect on politics, 36; explanation for, 12, 17–19, 29, 113–15, 273–76, 319n19; group medical care model, 278–79; income from stocks, 320n42; industry-­ specific reforms, 277; innate differences and, 113; licensing rule changes, 279; market egalitarianism, 289–90; merit-­based, 19–20; patent reform and, 281; presidential election and, 37–38; racial diversity and, 45; software developers and, 27–28; STEM workers and, 26–27; trade protectionism and, 278. See also top one p­ ercent Infant Health and Development Program, 132

374 • Index innovation, 29, 115, 274–75, 287, 296, 315; African Americans and, 184, 194, 283; blocking of, 35; in financial ser­v ices, 31; healthcare, 278; history of, 124, 163, 165–66; trade and, 103–4, 278 innovation, global, 105–7 innovation, Soviet Union, 65 Institute of Medicine, 246 insurance model of healthcare, 253–55, 278–79; history of, 252 intellectual property, 103–4, 106, 281 International ­Labor Organ­ization, 14 IQ: Asian American success and, 153–54, 343n18; biological dysgenic effects, 126–27; crystalized or fluid intelligence, 330n9; early childhood intervention, 132; education and, 118–19, 122–­25; environment and, 88–90; evolutionary ideas and, 149–50; gender and, 86, 332n27; immigration and, 134–­36; income and, 73–74, 77, 79; job per­for­mance and, 91–92, 332n34; mea­sure­ment of, 330n8; merit pay and, 75; military draftees tested (1919), 168–71; race, religion and, 81, 85, 129–31, 151–53, 168–75, 292, 338n39. See also ge­ne­tics; twin studies Jaffe, Sonia, 34 Jardina, Ashley, 48 Jefferson, Thomas, 148 Jensen, Arthur, 152–53 Jewish success theories, 153–58, 344n45, 347n86; discrimination and achievements and, 154–56, 343n23; ­family size theory and, 180; flaws in, 158–60; ge­ne­tic argument, 154; IQ scores and, 152, 168–75, 346–47n81; literacy and education theory and, 156–57; parents’ education and, 184–85; professions and occupations and, 160–­62, 163, 166; selective breeding theory and, 156; study habits and, 188 Jim Crow, 269; African Americans in northern states, 158, 169, 173;

coordinated exclusion against black Americans, 198; government participation in, 200; housing segregation and, 202, 212; origin of term, 197–98; overt prejudice of, 43, 51; police encounters and, 235; realtor’s code of ethics ­toward, 202 Jones, Charles, 306 Joshua ben Gamla, 157 justice: Aristotle’s writings, 52; balanced scales, 326n3 just society: communist view of, 64–65; definition of, 99; government, role of, 63; libertarian view of, 65–67, 100; market access in, 100–102; Plato’s view of, 59, 61, 66, 328n40; Rawls difference princi­ple of, 62–64; Rawls’ princi­ples of, 61–67; in Sparta, 328n40. See also income in­equality Katz, Larry, 220, 226 Kemper, D., 196 Klenow, Peter, 306 Kong, Augustine, 144 Kteily, Nour, 48–49 ­labor and ­labor ­unions: decline of, 13; pensions of, 14; share of income and, 14, 319n19 Latimer, Lewis Howard, 164 Laws (Plato), 60–61, 63 lead exposure, 227–28 Lee, James J., 142, 144 ­legal profession: restrictions on, 259; unmet ­legal needs in the, 263–64; Washington State ­limited practice officers, 264–66. See also American Bar Association (ABA) ­legal profession, global, 258–­60 Leibrand, Harry, 199 Lemley, Mark, 281 Le Pen, Marine, 1 Levinson, Boris, 175 Lewis, William H., 101 ­Limited License ­Legal Technician (LLLT), 265–66 Lindsey, Brink, 275

Index • 375 literacy and numeracy: Industrial Revolution and, 181–82; mindset and learning, 189–90; reading to ­children, 187, 349n115; society-­wide forces, 184–86 Logan, Trevon, 184 Loos, Clifford, 252 Luxembourg Income Study (LIS): executive earnings, 322n66; ­legal professionals, 258, 323n73; medical and l­ egal professionals, 28; STEM workers, 323n74; top one-­percent earners global, 20, 296, 301, 303, 313, 321n59, 322n61 Lynn, Richard, 155, 159, 167–68, 170–72 Mankiw, Greg, 18–19 Manning, Alan, 71 manufacturing: employment in, 9–10, 13, 41; tariffs and, 105; top earners in, 29; top one p­ ercent and, 12 manufacturing, global, 21–23 Martin, Trayvon, 235 Marx, Karl, 64 Massey, Douglas S., 110–11, 192, 202, 221, 229 Mather, Cotton, 166 Mazzucato, Mariana, 106 McCloskey, Deirdre, 181 McCoy, Elijah, 164 McDaniel, Michael, 92 McElwee, Sean, 48 McKay, Patrick, 92, 94 Medical Society of the District of Columbia, 253 Menand, Louis, 149 Miano, Armando, 47 Mishel, Larry, 15 Mokyr, Joel, 163, 166 monkey experiments, 54–55 monopolistic practices, 243–44; banks, 268; elite professionals and, 279, 292–93; historical, 100; ­lawyers, 259, 262, 265; patent rules and, 276 Montgomery, Benjamin T., 166 Morrill Act, 182 Morris, Mark, 94

Moving to Opportunity (MTO), 225–26 Murray, Charles, 81, 118–19, 152–53, 155–58, 168, 180 Musterd, Sako, 216 National Academies of Sciences, Engineering, and Medicine, 11, 12, 168, 246 National Assessment of Education Pro­g ress, 122 National Bureau of Asian Research, 103 National Center for State Courts, 265 National Front (France), 1–2 National Inventors Hall of Fame, 163–64 nationalism, definition of, 9 nationalist parties or politicians, global, 1–2, 39; racial and cultural differences in, 43–­4 4 National Longitudinal Survey of Youth (NLSY), 73; black-­white disparity, 81, 84–85, 331n19; Family/Home Risk Index, 87–88; health status, 75; IQ and religion, 172; mea­sure­ments, 330nn8, 10; social status of parents, 185; tele­vi­sion and homework time, 187 National Survey on Drug Use and Health, 238 neighborhood effect, 221, 223–24 neighborhoods: crime victimization in, 233–34; effect on health and prosperity, 219; exposure to criminality in, 220; government ser­vices in, 219; housing the poor, studies of, 225–27; immigrants and immigration and, 223; Israeli study, 224–25; life outcomes and, 221–23, 229, 355n27; Mount Laurel affordable housing and, 229; police encounters in, 234–36; safety and security in, 229–­32; Swedish study, 224; UK public housing study, 226 Newcomb, Edward J., 199 Newport, Frank, 42 news and media sources, 276–77 New York City Zoning Resolution, 206 New York State Combined Retirement System, 14 New York State ­Mental Deficiency Law, 150 New York Times, 14, 196

376 • Index Nisbett, Richard, 140 noncognitive skills, 18, 72, 87, 97, 128, 243, 292 North Carolina Board of Dental Examiners, 258 Northrup, Solomon, 166 Norway: cognitive per­for­mance in, 126–27; ­legal professionals in, 79 Nozick, Robert, 66 nurse prac­ti­tion­ers: efficacy of, 246–48; in­de­pen­dence for, 249; payments to, 245; restrictions on, 247; training, 245–46 nurse prac­ti­tion­ers, Canada, 256 nursing and allied professions, 244–47; maternal care and, 132; po­liti­cal opposition ­toward, 247–49; salaries of, 97 Nussbaum, Martha, 66 O*NET program, 85, 331n20 Obama, Barack, 42, 276 Oh, In-­Sue, 91 Okbay, Aysu, 142 Olson, Mancur, 275 “On Duties” (Cicero), 53, 59 Onesimus, 166 Organisation for Economic Co-­Operation and Development (OECD): capital income and, 14–15; economic slowdown, 4; executives and man­ag­ers, 24–26; global IQ study, 86, 332n26; government confidence, 2–­3, 7–8; immigration and trade, effect on, 12; ­lawyers’ mono­poly, 259; nationalist statements in, 43; politics, 43; price indexes, 33; top earners in, 31; top earners job sectors, 22–24. See also Programme for International Student Assessment (PISA); Programme for the International Assessment of Adult Competencies (PIAAC) Organisation for Economic Co-­Operation and Development (OECD) countries: gender and education, 332n27; healthcare spending, 29; income in­equality, 306, 311, 313; monopolistic practices, 243–44; neighborhood

satisfaction, 211; nonlabor income, 303–4; predictors of in­equality, 296; teacher salaries, 285; top earners in, 20 Our Kids (Putnam), 117 Page, Benjamin, 36–37 Painter, Nell Irvin, 149 Palmer, Maxwell, 208 parochialism, 58 Parton, Dolly, 187 patents, 124; African Americans and, 163–66; filing fees and, 259; healthcare and, 276; reforms of, 281 Peabody vocabulary IQ test, 338n39 Pendall, Rolf, 203–4 Penn World T ­ able, 303 Perry, Andre, 191 personality assessments, 87, 90, 172 personality development, 88 personality traits, 73–­76, 79, 81, 87–88, 91–92, 132, 332n34; entrepreneurship and, 96; ge­ne­tic advantages and, 113. See also cognitive ability; conscientiousness Pew Research Center, 46 Phar­ma­ceu­ti­cal Research and Manufacturers Association of Amer­i­ca, 251 Philipson, Tomas, 34 Phinney, Paul R., 249 physicians and surgeons, 25–26; educational requirements for, 256; entry barriers in, 245–­46; fee setting, 243; as homeowners, 301; insurance scrutiny of, 254–55; Jewish employment of, 160; limitations on practice by, 247–48; overprescription of treatment and tests, 278–79. See also American Medical Association (AMA) Physicians Foundation, 247–48 Piketty, Thomas, 5–7, 16, 296, 299 Pinker, Steven, 155 Piston, Spencer, 48 Plato, 59, 63–64, 66, 99, 106, 145, 148, 157; on education, 60, 157; on wisdom of rulers, 61; on ­women’s role, 60 Plomin, Robert, 138, 143 Polderman, Tinca, 139, 141

Index • 377 police: encounters with black men by, 234–36; segregation and, 234–36; stop-­and-­frisk policy of, 236–37 po­liti­cal equality, 98–100, 112–14, 117, 167, 182–84, 274, 278, 289, 293 po­liti­cal Left: pay in­equality, 69, 75; suspicion of power, 289 po­liti­cal Right: distrust of elites, 307; pay in­equality, 69, 75 politics: adult education, 333n1; ­labor market in­equality, 97; po­liti­cal equality, 99–100; presidential election and, 39–42; racial in­equality and, 39, 41–43, 324n95; top one ­percent role and, 36–37 pop­u­lism, 9–10 Practice of Law Board (Washington state), 264–65 predictors of income, 330n12, 330n13, 330n15; cognitive ability as, 73–75, 77–80, 82, 84, 91–92; educational attainment as, 73, 75, 77; health status as, 73, 75, 77; personality traits as, 73–74, 96 Pren, Karen, 111 Prince­ton University, 101 productivity: definition of, 69; evaluation of, 69–70 productivity growth, 4; advanced industries and, 31, 323n79; definition of, 29–31; effect on politics, 36; explanation for slowdown of, 29; innovation and, 105; living standards and, 306; mea­sure­ment of, 323n78; multifactor productivity, 323n76; real GDP and employment data, 323n81; rising returns to capital, 323n80; slowdown of, 31, 33–35, 37, 274 professional associations, 242–43 Programme for International Student Assessment (PISA): Asian test scores, 177–78; country of birth data, 137, 337n31; immigrant test scores, 137; trends in math scores, 126; ­women’s pro­g ress, 86–87 Programme for the International Assessment of Adult Competencies (PIAAC), 79, 84, 330n11, 331n19; cognitive ability

testing, 76–77, 322n71; cognitive ability test scores, 177–79; immigration status, 135; IQ of immigrants, 135–36; personality traits in, 77; racial and ethnic differences, 81, 331n17 Progressive Policy Institute, 276 Proj­ect Manifesto, 43 Proj­ect Talent, 86, 122, 123, 172–73, 180, 347n87 Putnam, Robert, 117 Putterman, Louis, 45 race and ethnicity: business owner­ship and, 96; employee engagement and, 92–­95; eugenics movement and, 150; fair treatment and, 94–96; immigration and, 107–8; noncognitive skills and, 87–88; per­for­mance evaluations and, 92; STEM knowledge and, 85; test per­for­mances and, 81, 84–85. See also education, in­equality in; Jewish success theories racial diversity, global, 45, 274, 296, 310–14 racism, 45–47, 315; alt-­right and, 49; antiracism and, 281–83; blocked access to markets, 101; colleges blocked African Americans, 101; drugs, sale or use, 237–41, 357nn48, 49; ge­ne­tic theories and, 149–51; housing and, 200; housing segregation, 196; housing segregation and, 101–2, 192, 194; immigration and, 46–47; in­equality and, 274; in South Africa, 211; stop-­and-­frisk policy and, 236–37; white ­people’s views and, 48–49, 291. See also housing segregation Ravitch, Diane, 118 Rawls, John, 61–65, 66–67, 99–100 reciprocity, 52, 67, 104, 273, 289 Redbird, Beth, 277 Reeves, Richard, 17 regulation reform, 276–77; financial products and, 267–68; housing market and, 274; ­labor law and, 112; mortgage terms and, 203; quality of goods and, 102. See also physicians and surgeons; zoning laws Reich, Robert, 13 religion, employee engagement, 94–­95

378 • Index Republic, The (Plato), 60–61, 328n40 Republicans: ­middle income voters, 39; presidential election and, 37; racial bias of, 42–43; view of Trump, 38 Rhodes, Jason, 206 Rice, Thomas Dartmouth, 197–98 Ricoeur, Paul, 64 Rietveld, Cornelius, 141–42 Right and alt-­right: economic slowdown and, 39; princi­ples of, 9, 48–­49, 67; racial bias of, 47; racism of, 48–50, 151; top one ­percent theories of, 26; Trump supporters, 48. See also po­liti­cal Right Ritchie, Stuart J., 122 Rockoff, Jonah, 127 Rogeberg, Ole, 127 Rognlie, Matthew, 14 Romalis, John, 10 Rosenthal, Erich, 170 Ross, Cody, 236 Ross, Donald, 252 Rothstein, Richard, 192 Royal Dutch Dentistry Association, 257 Rugh, Jacob, 202 Rushton, Phillipe, 152–53, 168 Saez, Emanuel, 6–7, 16, 299 Sanders, Bernie, 9, 13, 16, 121 Schmidt, Frank, 91 Seawright, Jason, 36 Securities and Exchange Commission (SEC), 271 securities and investment industry: access to hedge, private equity and venture capital funds, 280; hedge funds and mutual funds, 269–72; retail banks or institutional investors, 267–69; top earners in, 266–67, 364n91 segregation: anti-­market policy and, 197–99; crime victimization and, 233–36; economic, 209–10, 216–17, 222–23, 231; effects of, 187; policy of, 207; racial, 193, 222–23, 282. See also housing segregation; lead exposure; zoning laws segregation, global, 210–16, 310; apartheid as, 211

selfish be­hav­ior, 54–56, 58 Sen, Amartya, 66 ser­vice industries, 24, 35, 275, 322n62 Shadid, Michael, 252–53 Shaffer, Jonathan, 91 Sharkey, Patrick, 221 Shelly v. Kramer (U.S.), 200 Sherman Anti-­Trust Act, 253 Sides, John, 43 single-­payer healthcare, 279 slavery: Aristotle’s theory of, 148–49; Cicero’s statement against, 326n10 Smith, Adam, 106 socially valuable ser­vices, 71 Social Science Ge­ne­tic Association Consortium, 143, 189 Society of American Law Teachers, 261 Socrates, 60 Splinter, David, 317n9 Stancheva, Stephanie, 47 Stanford-­Binet test, 73, 171 Stevens, Alexander H., 148 Study of Statewide Early Education Programs, 128 Summers, Larry, 86 Survey of Consumer Finances, 6, 318n12 Sutherland, George, 206 sympathy, 9, 42, 105, 273 Taleb, Nassim, 272 Tarbell, Ida, 276 tax policies, in­equality and, 13, 16–17, 320n30 teachers: biases of, 46; exploitation of, 66; salary inequalities of, 70–71, 97, 285; value-­added quality of, 121, 127–31, 337n16. See also CLASS technological innovation, 2; communications and, 19; finance and, 268; in­equality and, 273, 310; minor Jewish role in, 163; post Industrial Revolution, 34–35; top one ­percent and, 26 Teles, Steven, 275 Tesler, Michael, 43 Texas Ophthalmologist Association (TOA), 250 Toner, Phillip, 287

Index • 379 top one ­percent: arts, design, entertainment workers, 20–21; data analy­sis, 6, 295–305, 308–­9 ; executives and man­a g­ers, 25–26, 322n67; global income share, 5; Japan, 322n62; job sectors of, 22–24, 26–27, 321n59, 364n90; ­legal profession, 24, 26, 28; occupation or sector, 21, 27; physicians and surgeons, 25–26, 28, 243–45; STEM workers, 28; trade-­oriented sectors, 23–24 trade: hunter-­gather tribes, 328n31; job loss and, 10 Trounstine, Jessica, 208 Trump, Donald, 1, 7, 9, 38–39, 41–43, 47–48, 50, 134 Tucker-­Drob, Elliot, 122 Turkheimer, Eric, 140 Twelve Years a Slave (Northrup), 166 twin studies: educational attainment and, 140; environmental effects and, 141, 341n68; IQ heritability and, 139–40, 340n59. See also ge­ne­tics

Wade, Nicholas, 155–56, 158 Walker, C. J., 166 Wall Street Journal, 138, 153 War on Drugs, 239–­4 0 Washington, Booker T., 183 Washington Acad­emy of Surgery, 253 Washington Park Court Improvement Association, 198 Washington State Bar Association, 264 Wasserstein, Bernard, 159 Wedow, Robbee, 142 Weil, David, 45 Whitten, Robert, 206 Wilson, E. O., 58 Wissenschaftszentrum Berlin für Sozialforschung (WZB), 43 Wodtke, Geoffrey, 221 Wolff, Edward, 6 World Bank, 105, 146, 304, 313, 342n86 World Income Database, 322n61 World In­equality Database (WID), 296–98, 301, 303, 313, 321n46 World Trade Organ­ization, 22

U.S. Constitution: Bill of Rights, 63, 100; ­Fourteenth Amendment, 167, 200

“yellow jacket” movement, 2 Yerkes, Robert, 168

Vanguard, 269 Vavreck, Lynn, 43 ­Virginia Bar Association, 262 ­Virginia Beach School District, 287–88

Zimmerman, George, 235 zoning laws, 204–9, 281–83; Boston actions, 208; New York actions, 206 Zucman, Gabriel, 6–7, 16, 299